This is a design-science paper about methods for explaining data-driven classifications of text documents. Document classification has widespread applications, such as with web pages for advertising, emails for legal discovery, blog entries for sentiment analysis, and many more. Document data are characterized by very high dimensionality, often with tens of thousands to millions of variables (words). Many applications require human understanding of the reasons for classification decisions: by managers, client-facing employees, and the technical team. Unfortunately, due to the high dimensionality, understanding the decisions made by the document classifiers is very difficult. Previous approaches to gain insight into black-box models do not deal well with high-dimensional data. Our main theoretical contribution is to define a new sort of explanation, tailored to the business needs of document classification and able to cope with the associated technical constraints. Specifically, an explanation is defined as a set of words (terms, more generally) such that removing all words within this set from the document changes the predicted class from the class of interest. We present an algorithm to find such explanations, as well as a framework to assess such an algorithm's performance. We demonstrate the value of the new approach with a case study from a real-world document classification task: classifying web pages as containing adult content, with the goal of allowing advertisers to choose not to have their ads appear there. We present a further empirical demonstration on news-story topic classification using the 20 Newsgroups benchmark dataset. The results show the explanations to be concise and document-specific, and to provide insight into the exact reasons for the classification decisions, into the workings of the classification models, and into the business application itself. We also illustrate how explaining documents' classifications can help to improve data quality and model performance.

Since the concept of crowdsourcing is relatively new, many potential participants have questions about the AMT marketplace. For example, a common set of questions that pop up in an 'introduction to crowdsourcing and AMT' session are the following: What type of tasks can be completed in the marketplace? How much does it cost? How fast can I get results back? How big is the AMT marketplace? The answers for these questions remain largely anecdotal and based on personal observations and experiences. To understand better what types of tasks are being completed today using crowdsourcing techniques, we started collecting data about the AMT marketplace. We present a preliminary analysis of the dataset and provide directions for interesting future research.

We present a model of dynamic monopoly pricing for a good that displays network effects. In contrast with the standard notion of a rational-expectations equilibrium, we model consumers as boundedly rational, and unable either to pay immediate attention to each price change, or to make accurate forecasts of the adoption of the network good. Our analysis shows that the seller's optimal price trajectory has the following structure: the price is low when the user base is below a target level, is high when the user base is above the target, and is set to keep user base stationary once the target level has been attained. We show that this pricing policy is robust to a number of extensions, which include the product's user base evolving over time, and consumers basing their choices on a mixture of a myopic and a "stubborn" expectation of adoption. Our results differ significantly from those that would be predicted by a model based on rational-expectations equilibrium, and are more consistent with the pricing of network goods observed in practice.

We define an economic network as a linked set of entities, where links are created by actual realizations of shared economic outcomes between entities. Such networks are becoming increasingly prevalent on the Internet, an example being the copurchase netwok on Amazon where entities are books and links designate which pairs were purchased simultaneously. Our dataset covers a diverse set of books spanning over 400 categories over a period of three years with a total of over 70 million observations. To our knowledge, this is the first large scale study showing that an economic network contains useful predictive information that is distributed in the network. We show that an economic network contains predictive information. Specifically, we demonstrate that an entity’s future demand is more accurately predicted by combining its historical demand with that of its neighbors than by considering its demand alone. In other words, if you want to know what your state will be in the future, consider what is happening to your neighbors now. This result could apply to other economic networks where outcomes of sets of entities tend to be related.

Predictive models in regression and classification problems typically have a single model that covers most, if not all, cases in the data. At the opposite end of the spectrum is a collection of models each of which covers a very small subset of the decision space. These are referred to as “small disjuncts.” The tradeoffs between the two types of models have been well documented. Single models, especially linear ones, are easy to interpret and explain. In contrast, small disjuncts do not provide as clean or as simple an interpretation of the data, and have been shown by several researchers to be responsible for a disproportionately large number of errors when applied to out of sample data. This research provides a counterpoint, demonstrating that “simple” small disjuncts provide a credible model for financial market prediction, a problem with a high degree of noise. A related novel contribution of this paper is a simple method for measuring the “yield” of a learning system, which is the percentage of in sample performance that the learned model can be expected to realize on out-of-sample data. Curiously, such a measure is missing from the literature on regression learning algorithms.

We consider pure-jump transaction-level models for asset prices in continuous time, driven by point processes. In a bivariate model that admits cointegration, we allow for time deformations to account for such e®ects as intraday seasonal patterns in volatility, and non-trading periods that may be di®erent for the two assets. Most assumptions are stated directly on the point process, though we provide su±cient conditions on the corresponding inter-trade durations for these assumptions to hold. We obtain the asymptotic distribution of the log-price process. We also obtain the asymptotic distribution of the ordinary least-squares estimator of the cointegrat- ing parameter based on data sampled from an equally-spaced discretization of calendar time, in the case of weak fractional cointegration. Finally, we obtain the limiting distribution of the ordinary least-squares estimator of the autoregressive parameter in a simpli¯ed transaction-level univariate model with a unit root.

This paper demonstrates that "social network collaborative filtering" (SNCF), wherein user-selected like-minded alters are used to make predictions, can rival traditional user-to-user collaborative filtering (CF) in predictive accuracy. Us-ing a unique data set from an online community where users rated items and also created social networking links specifically intended to represent like-minded “allies,” we use SNCF and traditional CF to predict ratings by net-worked users. We find that SNCF using generic "friend" alters is moderately worse than the better CF techniques, but outperforms benchmarks such as by-item or by-user average rating; generic friends often are not like-minded. However, SNCF using "ally" alters is competitive with CF. These results are significant because SNCF is tremendously more computationally efficient than traditional user-user CF and may be implemented in large-scale web commerce and social networking communities. It is notoriously difficult to distinguish the contributions of social influence (where allies influence users) and "social” selection (where users are simply effective at selecting like-minded people as their allies). Nonetheless, comparing similarity over time, we do show no evi-dence of strong social influence among allies or friends.

With the rapid growth of the Internet, the ability of users to create and publish content has created active electronic communities that provide a wealth of product information. However, the high volume of reviews that are typically published for a single product makes harder for individuals as well as manufacturers to locate the best reviews and understand the true underlying quality of a product. In this paper, we re-examine the impact of reviews on economic outcomes like product sales and see how different factors affect social outcomes like the extent of their perceived usefulness. Our approach explores multiple aspects of review text, such as lexical, grammatical, semantic, and stylistic levels to identify important text-based features. In addition, we also examine multiple reviewer-level features such as average usefulness of past reviews and the self-disclosed identity measures of reviewers that are displayed next to a review. Our econometric analysis reveals that the extent of subjectivity, informativeness, readability, and linguistic correctness in reviews matters in influencing sales and perceived usefulness. Reviews that have a mixture of objective, and highly subjective sentences have a negative effect on product sales, compared to reviews that tend to include only subjective or only objective information. However, such reviews are considered more informative (or helpful) by the users. By using Random Forest based classifiers, we show that we can accurately predict the impact of reviews on sales and their perceived usefulness. Reviews for products that have received widely fluctuating reviews, also have reviews of widely fluctuating helpfulness. In particular, we find that highly detailed and readable reviews can have low helpfulness votes in cases when users tend to vote negatively not because they disapprove of the review quality but rather to convey their disapproval of the review polarity. We examine the relative importance of the three broad feature categories: `reviewer-related' features, `review subjectivity' features, and `review readability' features, and find that using any of the three feature sets results in a statistically equivalent performance as in the case of using all available features. This paper is the first study that integrates econometric, text mining, and predictive modeling techniques toward a more complete analysis of the information captured by user-generated online reviews in order to estimate their socio-economic impact. Our results can have implications for judicious design of opinion forums.

This paper describes an approach for incorporating externally specified aggregate ratings information into certain types of recommender systems, including two types of collaborating filtering and a hierarchical linear regression model. First, we present a framework for incorporating aggregate rating information and apply this framework to the aforementioned individual rating models. Then we formally show that this additional aggregate rating information provides more accurate recommendations of individual items to individual users. Further, we experimentally confirm this theoretical finding by demonstrating on several datasets that the aggregate rating information indeed leads to better predictions of unknown ratings. We also propose scalable methods for incorporating this aggregate information and test our approaches on large datasets. Finally, we demonstrate that the aggregate rating information can also be used as a solution to the cold start problem of recommender systems.

This paper addresses the repeated acquisition of labels for data items when the labeling is imperfect. We examine the improvement (or lack thereof) in data quality via repeated labeling, and focus especially on the improvement of training labels for supervised induction. With the outsourcing of small tasks becoming easier, for example via Rent-A-Coder or Amazon's Mechanical Turk, it often is possible to obtain less-than-expert labeling at low cost. With low-cost labeling, preparing the unlabeled part of the data can become considerably more expensive than labeling. We present repeated-labeling strategies of increasing complexity, and show several main results. (i) Repeated-labeling can improve label quality and model quality, but not always. (ii) When labels are noisy, repeated labeling can be preferable to single labeling even in the traditional setting where labels are not particularly cheap. (iii) As soon as the cost of processing the unlabeled data is not free, even the simple strategy of labeling everything multiple times can give considerable advantage. (iv) Repeatedly labeling a carefully chosen set of points is generally preferable, and we present a robust technique that combines different notions of uncertainty to select data points for which quality should be improved. The bottom line: the results show clearly that when labeling is not perfect, selective acquisition of multiple labels is a strategy that data miners should have in their repertoire; for certain label-quality/cost regimes, the benefit is substantial.

The growing pervasiveness of the Internet has changed the way that consumers shop for goods. Increasingly, user-generated product reviews serve as a valuable source of information for customers making product choices online. While there is a significant body of theory on multi-attribute choice under uncertainty, the literature that examines product reviews has not built on this stream of theory for a variety of reasons. Typically, the impact of product reviews has been incorporated by numeric variables representing the valence and volume of reviews. In this paper we posit that the information embedded in product reviews cannot be captured by a single scalar value. Rather, we argue that product reviews are multifaceted and hence, the textual content of product reviews is an important determinant of consumers' choices, over and above the valence and volume of reviews. We provide a text mining technique that allows us to incorporate text in choice and panel data models by decomposing textual reviews into segments, evaluating different product features. We test our approach on a unique dataset collected from Amazon, and demonstrate how it can be used to learn consumers' relative preferences for different product features. The dataset used contains three different groups of products (digital cameras, camcorders, PDAs), associated sales data and consumer review data gathered over a 15-month period. Additionally, we present and discuss two experimental techniques that can be used to alleviate the problem of data sparsity and of omitted variables: the first technique models consumer opinions as elements of a tensor product of independent feature and evaluation spaces and the second technique clusters rare opinions based on pointwise mutual information. The paper concludes by discussing the managerial relevance of this work as a tool for extracting actionable business intelligence from user-generated content.

The paper presents a method that uses aggregate ratings provided by various segments of users for various categories of items to derive better estimations of unknown individual ratings. This is achieved by converting the aggregate ratings into constraints on the parameters of a rating estimation model presented in the paper. The paper also demonstrates theoretically that these additional constraints reduce rating estimation errors resulting in better rating predictions.

We develop a formal model of online-offline substitution to identify three factors that drive consumers to purchase online: convenience, selection, and price. This model builds hypotheses on how features of offline retail supply impact online purchasing. We then examine how the local availability of offline retail options drives use of the online channel and consequently how the convenience, selection, and price advantages of the online channel may vary by geographic location. In particular, we examine the effect of local store openings on online book purchases in that location. We explore this problem using data from Amazon on the top selling books for 1501 unique locations in the US for 10 months ending in January 2006. In addition to this data, we use information on changes in local retail competition as measured by openings of large specialty bookstores such as Borders or Barnes & Noble and discount stores such as Wal-Mart or Target. We show that even controlling for product-specific preferences by location, changes in local retail options have substantial effects on online purchases. We demonstrate how the convenience, selection, and price benefits of the Internet are different for customers in different types of locations. More generally, we show that geography significantly impacts the benefit that consumers derive from electronic markets.

The Sarbanes Oxley Act (SOA) introduced significant changes to financial practice and corporate governance regulation, including stringent new rules designed to protect investors by improving the accuracy and reliability of corporate disclosures. Briefly speaking, it requires management to submit a report containing an assessment of the effectiveness of the internal control structure, a description of material weaknesses in such internal controls and of any material noncompliance. Such mandatory regulations can have some broader ramifications on firm profitability, market structure and social welfare, many of which were unintended when policy makers first formulated this Act. Moreover, the tight coupling between compliance activities, information disclosure and IT investments can have implications for IT governance because of its potential to change relationships between technology investments and business. This article aims to provide some intuitive insights into the trade-offs involved for firms in disclosure of such information, and gives an overview of some research questions that would be of interest to academics, industry executives and policy makers alike.

Three of the most important uses of the Internet today are as an economic marketplace, as a forum for social interaction, and as a source of information. In this paper, we explore how these three activities come together, in the form of emergent social communities built around information exchanges within IT-enabled electronic marketplaces. Drawing on social identity theory, we suggest that the relationship between online consumer reviews and internet product sales is partially explained by social identity processes. Using a unique dataset based on both chronologically compiled ratings as well as reviewer characteristics for a given set of products and geographical location-based purchasing behavior from Amazon, we provide evidence at the community level linking the prevalence of identity claiming behavior in an online community with subsequent product sales. In addition, we show that when reviewers claim to be from a particular geographic location, subsequent product sales are higher in that region. At the review level of analysis, we show that subsequent reviews conform to identity-claiming norms set in previous reviews, and that identity claiming that conforms to community norms elicits identity granting. Furthermore, our results suggest that the prevalence of identity granting has implications for economic exchange in the form of product sales. Implications for research on word-of-mouth and electronic communities are discussed.

Large amounts of (often valuable) information are stored in web-accessible text databases. ``Metasearchers'' provide unified interfaces to query multiple such databases at once. For efficiency, metasearchers rely on succinct statistical summaries of the database contents to select the best databases for each query. So far, database selection research has largely assumed that databases are static, so the associated statistical summaries do not need to change over time. However, databases are rarely static and the statistical summaries that describe their contents need to be updated periodically to reflect content changes. In this article, we first report the results of a study showing how the content summaries of 152 real web databases evolved over a period of 52 weeks. Then, we show how to use ``survival analysis'' techniques in general, and Cox's proportional hazards regression in particular, to model database changes over time and predict when we should update each content summary. Finally, we exploit our change model to devise update schedules that keep the summaries up to date by contacting databases only when needed, and then we evaluate the quality of our schedules experimentally over real web databases.

Often, in the real world, entities have two or more representations in databases. Duplicate records do not share a common key and/or they contain errors that make duplicate matching a difficult task. Errors are introduced as the result of transcription errors, incomplete information, lack of standard formats or any combination of these factors. In this article, we present a thorough analysis of the literature on duplicate record detection. We cover similarity metrics that are commonly used to detect similar field entries, and we present an extensive set of duplicate detection algorithms that can detect approximately duplicate records in a database. We also cover multiple techniques for improving the efficiency and scalability of approximate duplicate detection algorithms. We conclude with a coverage of existing tools and with a brief discussion of the big open problems in the area.

We present a model of dynamic monopoly pricing for a good that displays network effects. In contrast with the standard notion of a rational-expectations equilibrium, we model consumers as boundedly rational, and unable either to pay immediate attention to each price change, or to make accurate forecasts of the adoption of the network good. Our analysis shows that the seller’s optimal price trajectory has the following simple structure: the price is zero when the product user base is below a specific threshold, and is chosen to keep user base stationary once this threshold demand level has been attained. We show that our prescribed pricing policy is robust to a number of extensions, which include the product’s user base evolving over time, a fraction of consumers being sufficiently rational to make accurate adoption forecasts, and consumers basing their choices on a mixture of a myopic and a "stubborn" expectation of adoption. Our results differ significantly from those that would be predicted by a model based on rational-expectations equilibrium, and are more consistent with the pricing of network goods observed in practice.

The effective management of digital rights is the central challenge in many industries making the transition from physical to digital products. We present a new model that characterizes the value of these digital rights when products are sold both embedded in tangible physical artifacts, and as pure digital goods, and when granting rights permitted by one’s digital rights management (DRM) platform may affect the extent of digital piracy. Our model indicates that in the absence of piracy, digital rights should be unrestricted, since a seller can use its pricing strategy to optimally balance sales between physical and digital goods. However, the threat of piracy limits the extent to which digital rights should be granted: the value of digital rights is determined not only by their direct effect on the quality of legal digital goods, but by a differential piracy effect that can lower a seller’s pricing power. When the latter effect is sufficiently high, granting digital rights can have a detrimental effect on value — our model indicates that this kind of effect is more likely to be observed for digital rights that aim to replicate the consumption experience of physical goods, rather than enhancing a customer’s digital experience. We test the predictions of our analytical model using data from the ebook industry. Our empirical evidence supports our theoretical results, showing that four separate digital rights each have an economically significant impact on ebook prices, and establishing that the digital rights which aim to replicate physical consumption while increasing the threat of piracy are the ones that have negative impact on seller value. We also show that if the pricing of a digital good is keyed off that of an existing tangible good, optimal pricing changes for the former should be more nuanced, rather than simply mirroring changes in the price of the latter, and we discuss the effect of the technological sophistication of potential customers on optimal pricing and rights management. Our results represent new evidence of the importance of an informed and judicious choice of the different digital rights granted by a DRM platform, and provide a new framework for guiding managers in industries that are progressively being digitized.

Positive externalities characterize the consumption of a majority of information goods such as software, various Internet services, and online communities. In a simple model of vertical differentiation, we show that network externality is a critical factor for the versioning of such information goods. In particular, a multi-product monopolist offers two versions of distinct qualities. The underlying rationale is that offering the low-end version expands the network size and thus enhances the (network) value of the high-end version, allowing the firm to charge a higher price for the high-end version. In addition, we show that the low-quality version may be offered for free under very general conditions. Competition between firms producing compatible products reduces their incentive to version their products due to the spillover effects in a shared product network.

Information Systems Working Papers Series

We investigate the hypothesis: those consumers who have communicated with a customer of a particular service have increased likelihood of adopting the service. We survey the diverse literature on such "viral marketing," providing a categorization of the specific research questions asked, the data analyzed, and the statistical methods used. We highlight a striking gap in the literature: no prior study has had both of the two key types of data necessary to provide direct support for the hypothesis: data on communications between consumers, and data on product adoption. We suggest a type of service for which both types of data are available telecommunications services. Then, for a particular telecommunication service, we show support for the hypothesis. Specifically, we show three main results. 1) there is such a "viral" effect and it is statistically significant, resulting in take rates 3-5 times greater than a baseline group; 2) attributes constructed from the consumer network can improve models for ranking of targeted customers by likelihood of adoption, and 3) observing the network allows the firm to target new customers that would have fallen through the cracks, because they would not have been identified based solely on the traditional set of attributes used for marketing by the firm. We close with a discussion of challenges and opportunities for research in this area. For example, can one determine whether the reason for the viral effect is customer advocacy (e.g., via "word of mouth") versus network-identified homophily?

The extant literature on price promotions typically assumes that consumers loyal to a brand never switch to a competing brand, with Shilony (1977) and Raju et al (1990) being exceptions. Extending the Narasimhan (1988) model, we allow loyal consumers to hold finite brand loyalty. Our unique equilibrium splits into three types, depending upon configurations of consumer reservation utility, brand strength and switcher population. The type of equilibrium for high brand loyalty corresponds to the one in Narasimhan (1988). The remaining two types for intermediate and low brand loyalty demonstrate strikingly different properties. First, the strong brand has a higher price range and a higher regular price. Second, the strong brand has a higher (lower) average promotional depth than the weak brand when the switcher population is small (large). Third, both brands promote equally frequently when brand loyalty is relatively low. Therefore, this analysis hopefully provides a more complete picture about firms’ promotional decisions for all possible levels of brand loyalty and switcher pupulation.

Initially popularized by Amazon.com, recommendation technologies have become widespread over the past several years, both in the industry and academia. The traditional two-dimensional approach to recommender systems, involving the dimensions of Users and Items, has been subsequently extended to the multidimensional approach supporting additional contextual dimensions and OLAP-type aggregation capabilities. Furthermore, the class of all possible recommendations available to the users in traditional recommender systems is typically determined by the vendor and is quite limited. In this paper we address this limitation by proposing a query language RQL that allows the users to formulate various types of recommendation requests on their own. RQL adapts OLAP queries to the domain of recommender systems and, therefore, is able to support both the traditional two-dimensional and the more complex multidimensional recommender systems. The paper also presents a recommendation algebra that allows mapping RQL queries into the algebraic expressions for the query processing purposes. Finally, the paper presents a method for executing RQL queries.

This paper is about constructing confidence bands around an ROC curve such that (1 - \delta)% of the ROC curves traced by data sets of size r will fall completely within the bands. We introduce to the machine learning community three methods from the medical field that are applicable to generate such bands. We then evaluate these methods on the simple case of “binormal” distributions— the scores for positive and the score for negative instances are drawn from two normal distributions. We show that none of the methods generate appropriate bands and investigate two types of variances problems. We show that widening the bands does not produce the proper bandwidths but that fitting a normal distribution to the observed drawn samples and drawing samples from this distribution (parametric bootstrap) does generate bands that are much closer to the desired coverage although still not perfect. We tested the original methods as well as parametric bootstrap on the covertype data set from the UCI ML-repority. The original methods perform the same as in the synthetic case, whereas the parametric bootstrap technique did not yield the expected results. This is primarily due to not being able to generate a good fit for the score distributions. Whether it is possible to fit well-behaving parametric distribution to learned models is an open question we leave to the machine learning community to answer.

We analyze a revenue management problem in which a seller facing a Poisson arriving stream of customers operates an online multiunit auction. Customers have an alternative list price channel where to get the product from. We consider two variants of this problem: In the first one, the list price is an external channel run by another firm. In the second variant, the seller manages simultaneously both the auction and the list price channels. Each consumer, trying to maximize his own surplus, must decide either to buy at the posted price and get the item at no risk, or to join the auction and wait until its end, where the winners are revealed and the auction price is disclosed. Our approach consists of two parts. First, we study structural properties of the problem, and show that the equilibrium strategy for both versions of this game is of the threshold type, meaning that a consumer will join the auction only if his arrival time is above a function of his own valuation. This consumer’s strategy can be computed using an iterative algorithm in a function space, provably convergent under some conditions. Unfortunately, this procedure is computationally intensive. To overcome this, we formulate an asymptotic version of the problem, in which the demand rate and the initial number of units grow proportionally large. We get a simple closed form for the equilibrium strategy in this regime, which is then used as an approximated solution for the original problem. Numerical computations show that this heuristic is very accurate. The asymptotic solution culminates then in simple and precise recipes for how bidders should behave, and how the seller should structure the auction, and price the product in the dual channel case.

An important simplifying assumption made when analyzing goods that display positive network effects is that potential consumers can form a rational expectation of the equilibrium demand for the good, and that they all form the same expectation, which is then fulfilled based on their consumption choices - sometimes called a rational expectations equilibrium (REE). We examine whether the results of these models are robust to the relaxation of this assumption. In our model, consumers differ in their marginal utility of total demand (intensity of the network effect), which varies according to a given distribution (the distribution of consumer "types"), and are boundedly rational in two ways. First, only a fraction of consumers "pay attention" to price announcements in any interval of time. Second, those consumers who pay attention make their consumption choices based on a boundedly rational expectation of future demand. Our benchmark model is of myopic expectations, although we show how our results generalize (1) to a case in which the population of consumers contains both those who are myopic and those who are "fully rational," and (2) to a case in which consumers have expectations that are partly "stubborn". We base our analysis on a continuous-time approximation of an underlying discrete-time model. Under this approximation, the instantaneous choices of consumers continuously influence the rate at which demand adjusts over time, and a monopolist chooses a price trajectory to maximize profit. First, we show that, under fairly general assumptions about the distribution of types, the profit-maximizing rational expectations equilibrium is not a steady state of the optimal trajectory with boundedly rational consumers. Our second theorem shows that if consumer types are uniformly distributed and consumers form myopic (or more rational) expectations, the monopolist’s optimal pricing trajectory is generated by a "target policy" with the following properties: when current demand is below the target, the price is zero; when current demand is above the target, the price is the maximum possible, and when current demand is at the target, the price is chosen to keep demand stationary. We also show that the optimal demand target with boundedly rational consumers is always strictly lower than the equilibrium level of demand predicted by a model with rational expectations. Furthermore, the difference between the target demand and the rational expectations demand is higher when consumers pay attention to the monopolist’s price announcements at a lower rate. We generalize the results from this example in two ways. Our third theorem examines the case of myopic consumers and strictly concave distributions of consumer types. To find an optimal policy one must expand the set of controls to include measure-valued controls. The optimal policy is similar to the target policy of Theorem 2, except that when current demand is at the target, the monopolist chooses the "mixture" between a price of zero and the maximum possible price that keeps demand stationary. For convex consumer type distributions in the neighborhood of the uniform distribution, we give a heuristic argument to support a conjecture that the monopolist continues to choose a demand target lower than the rational expectations demand, but varies price gradually in the neighborhood of the demand target. Finally, for uniformly distributed types and consumer expectations that are both myopic and "stubborn", we show that the monopolist’s optimal pricing trajectory is generated by a target policy with the same properties as those in Theorem 2, although with a target that is strictly lower, and that increases as consumers become progressively less stubborn. (This paper is part of a program of research whose broad objective is to explore the conditions under which the assumption of unbounded rationality in economic models is a reasonable one).

Internet Referral Services, hosted either by independent third-party infomediaries or by manufacturers serve as “lead-generators” in electronic marketplaces, directing consumer traffic to particular retailers. In a model of price dispersion with mixed strategy equilibria, we investigate the competitive implications of these institutions on retailer and manufacturer pricing strategies as well as their impact on channel structures and distribution of profits. Offline, retailers face a higher customer acquisition cost. In return, they can engage in price discrimination. Online, they save on the acquisition costs, but lose the ability to price discriminate. This critical tradeoff drives firms’ equilibrium strategies. The establishment of a referral service is a strategic decision by the manufacturer, in response to a third-party infomediary. It leads to an increase in channel profits and a reallocation of the increased surplus to the manufacturer, via the franchise fees. Further, it enables the manufacturer to respond to an infomediary, by giving itself a wider leeway to set the unit wholesale fee to the profit maximizing level. We discuss implications of referral services on channel coordination issues, and whether a two part tariff can be successfully used to maximize channel profits. Contrary to prior literature, we find that when retailers can price discriminate among consumers, the manufacturer may not set the wholesale price to marginal cost to coordinate the channel. Consistent with anecdotal evidence, our model predicts that while it is optimal for an infomediary to enroll only one retailer, it is optimal for a manufacturer to enroll both retailers. Finally, our results show that under some circumstances, the manufacturer even benefits from the presence of the competing referral infomediary and hence, will not want to eliminate it.

Given that Information Technology (IT) security has emerged as an important issue in the last few years, the subject of security information sharing among firms, as a tool to minimize security breaches, has gained the interest of practitioners and academics. To promote the disclosure and sharing of cyber-security information among firms, the US federal government has encouraged the establishment of many industry based Information Sharing & Analysis Centers (ISACs) under Presidential Decision Directive 63. Sharing security vulnerabilities and technological solutions related to methods for preventing, detecting and correcting security breaches, is the fundamental goal of the ISACs. However, there are a number of interesting economic issues that will affect the achievement of this goal. Using game theory, we develop an analytical framework to investigate the competitive implications of sharing security information and investments in security technologies. We find that security technology investments and security information sharing act as “strategic complements” in equilibrium. Our results suggest that information sharing is more valuable when product substitutability is higher, implying that such sharing alliances yield greater benefits in more competitive industries. We also highlight that the benefits from such information sharing alliances increase with the size of the firm. We compare the levels of information sharing and technology investments obtained when firms behave independently (Bertrand-Nash) to those selected by an ISAC which maximizes social welfare or joint industry profits. Our results help us predict the consequences of establishing organizations such as ISACs, CERT or InfraGard by the federal government.

This paper develops a game-theoretic model of "petty corruption" by gov- ernment officials. Such corruption is widespread, especially (but not only) in developing and transition economies. The model goes beyond the previ- ously published studies in the way it describes the structure of bureaucratic "tracks," and the information among the participants. Entrepreneurs apply, in sequence, to a "track" of two or more bureaucrats in a prescribed order for approval of their projects. Our first result establishes that in a one-shot situation no project ever gets approved. This result leads us to consider a repeated interaction setting. In that context we characterize in more detail the trigger-strategy equilibria that minimize the social loss due to the system of bribes, and those that maximize the expected total bribe income of the bureaucrats. The results are used to shed some light on two much advocated anti-corruption policies: the single window policy and rotation of bureaucrats.

There has been much discussion in the press about productivity improvements that grew at an annual rate of 3.55% from 2000 to 2003 [BW04]. One of the sources of this productivity growth is automation. We have all witnessed numerous ways in which companies have automated their business processes over the past decade. As a recent example, The Dallas Morning News reports in [Baj04] how Atmos Energy, the Dallas-based gas company, is automating its gas meter reading capabilities by using wireless technologies and thus reducing its staff by 225 employees over the next five years. In this article, we will examine current trends in the technology-driven automation and will argue that we are still in the early stages of a new wave of automation that will profoundly affect the economy and will significantly contribute to the productivity growth over the next 10 – 15 years. Industrial automation is an old phenomenon that goes back to the Industrial Revolution when machines replaced physical labor on a massive scale. Automation profoundly affected manufacturing over the past 25 years when industrial robots replaced various manual jobs in different spheres of manufacturing, including automobiles, computers and telecommunication equipment. More recently, automation was primarily driven by IT. For example, toll booth collectors recently became victims of IT-based automation when some of them lost their jobs to EZ-Pass technologies. Similarly, 225 employees at Atmos Energy will lose their jobs within the next 5 years due to the advancements in wireless technologies [Baj04]. Also, many cashiers in department stores and supermarkets will soon lose their jobs because of the advancements of the RFID tag technologies. Most of the jobs lost to automation have been routine production jobs, according to the job classification proposed by Robert Reich in [Rei91]. The main characteristics of these jobs are repetitiveness and structuredness since they have well defined procedural job descriptions. Examples of these jobs include assembly line workers, foremen, data processors, and toll collectors. The routine production jobs have been replaced by mechanical, electrical and IT-driven machines, including industrial robots and wireless communication devices. In this article, we claim that the next waive of automation will affect not only routine production workers, but also what Reich calls symbolic-analytic workers [Rei91], such as engineers, office and knowledge workers, managers, educators, and other groups of “mind workers.” Although few of these jobs will be eliminated completely, many of the more routine tasks in these jobs will be delegated to “smart machines” within the next 10 – 15 years, leading to major restructuring and consolidation of some of these jobs. This phenomenon is examined in the rest of this article.

Growth of business-to-consumer (B2C) applications such as electronic storefronts, catalogues, and customer support websites has drawn a great number of diverse stakeholders into the IS Development (ISD) practice. Marketing, strategy, and graphic design specialists have joined a variety of technical professionals and business stakeholders in developing B2C applications. Oftentimes, these professionals work for different organizations with different histories, cultures, and reward structures. A longitudinal qualitative field study of a B2C application development project was undertaken in order to build an in-depth understanding of the collaborative practices of diverse professionals in ISD projects. The paper proposes that the multi-party collaborative practice can be understood as a “collective reflection-in-action” cycle through which an IS design emerges as a result of agents producing, sharing, and reflecting upon material objects. Agents from diverse backgrounds exert different influences over emergent designs depending on their organization, profession, and project involvement-based power relations. These power relations shape whether collaborators “add to” “ignore,” or “challenge” the work produced by others. In turn, agents’ actions either reinforce or transform existing power relations depending on who gets to claim credit for and ownership of the emergent design. Implications for the study of boundary objects, team diversity, organizational learning, and contemporary ISD are drawn.

Knowledge Management (KM) literature has centrally focused on organization’s ability to build practices that integrate diverse expertise across professional, organizational, industry and other boundaries. In this paper we investigate how an organizational competence in boundary spanning emerges in practice. We draw on the concepts of boundary spanner and boundary object and on the practice-based view of KM in organizations to understand the emergence of boundary spanning in practice, which we define as relating practices from diverse fields. We contrast data from two qualitative, longitudinal field studies to draw our conclusions. We argue that for boundary spanning to emerge in practice a new joint field, which unites agent in a common pursuit, needs to be produced. Engagement of agents in this practice partially transforms their practices in local fields so as to accommodate the interests of their counterparts. Those agents who engage in negotiating the nature of this new field become boundary spanners-in-practice. Through their engagements in the new joint field and diverse local practices boundary spanners-in-practice produce and use objects which become locally useful and acquire a joint identity through their use – boundary objects-in-use. Through data analysis we find, first, that nominated boundary spanners and designated boundary objects do not always become boundary spanners-in-practice and boundary objects-in-use. Second, we outline the conditions necessary for boundary spanners-in-practice to emerge, including the need for them to become legitimate, albeit peripheral, participants in the practices of the fields that they span. Thirdly, we show how boundary spanners-in-practice use their symbolic, cultural, social, and economic resources (capital) to build the new joint field. Finally, we examine the tensions involved in a) the nomination of agents as boundary spanners and artifacts as boundary objects; b) the growth of the new joint field; c) agents’ choice in investing in the new joint field; and d) spanning one at the expense of another kind of boundary. We conclude by drawing implications for IS implementation and use.

There have been various claims made in the marketing community about the benefits of 1-to-1 marketing versus traditional customer segmentation approaches and how much they can improve understanding of customer behavior. However, few rigorous studies exist that systematically compare these approaches. In this paper, we conducted such a systematic study and compared the performance of aggregate, segmentation, and 1-to-1 marketing approaches across a broad range of experimental settings such as multiple segmentation levels, multiple real world marketing datasets, multiple dependent variables, different types of classifiers, different segmentation techniques, and different predictive measures. Our results show that, overall, 1-to-1 modeling significantly outperforms the aggregate approach among high-volume customers and is never worse than aggregate approach among low-volume customers in our experimental settings. Moreover, the best segmentation techniques tend to outperform 1-to-1 modeling among low-volume customers.

When entities are linked by explicit relations, classification methods that take advantage of the network can perform substantially better than methods that ignore the network. This paper argues that studies of relational classification in networked data should include simple network-only methods as baselines for comparison, in addition to the non-relational baselines that generally are used. In particular, comparing more complex algorithms with algorithms that only consider the network (and not the features of the entities) allows one to factor out the contribution of the network structure itself to the predictive power of the model. We examine several simple methods for network-only classification on previously used relational data sets, and show that they can perform remarkably well. The results demonstrate that the inclusion of network-only classifiers can shed new light on studies of relational learners.

The paper presents a survey of the field of recommender systems and describes current recommendation methods that are usually classified into the following three main categories: content-based, collaborative, and hybrid recommendation approaches. The paper also describes various limitations of current recommendation methods and discusses possible extensions that can improve recommendation capabilities. These extensions include, among others, improvement of understanding of users and items, incorporation of the contextual information into the recommendation process, support for multi-criteria ratings, and provision of more flexible and less intrusive types of recommendations.

This paper analyzes the optimal choice of pricing schedules and technological deterrence levels in a market with digital piracy, when legal sellers can sometimes control the extent of piracy by implementing digital rights management (DM) systems. It is shown that the seller's optimal pricing schedule can be characterized as a simple combination of the zero-piracy pricing schedule, and a piracy-indifferent pricing schedule which makes all customers indifferent between legal consumption and piracy. An increase in the level of piracy is shown to lower prices and profits, but may improve welfare by expanding the fraction of legal users and the volume of legal usage. In the absence of price-discrimination, the optimal level of technology-based protection against piracy is shown to be the technologically-maximal level, which maximizes the difference between the quality of the legal and pirated goods. However, when a seller can price-discriminate, it is always optimal for them to choose a strictly lower level of technology-based protection. Moreover, if a DRM system weakens over time, due to its technology being progressively hacked, the optimal strategic response may involve either increasing or decreasing the level of technology-based protection and the corresponding prices. This direction of change is related to whether the technology implementing each marginal reduction in piracy is increasingly less or more vulnerable to hacking. Pricing and technology choice guidelines based on these results are presented, and some social welfare issues are discussed.

Due to interest in social and economic networks, relational modeling is attracting increasing attention. The field of relational data mining/learning, which traditionally was dominated by logic-based approaches, has recently been extended by adapting learning methods such as naive Bayes, Baysian networks and decision trees to relational tasks. One aspect inherent to all methods of model induction from relational data is the construction of features through the aggregation of sets. The theoretical part of this work (1) presents an ontology of relational concepts of increasing complexity, (2) derives classes of aggregation operators that are needed to learn these concepts, and (3) classifies relational domains based on relational schema characteristics such as cardinality. We then present a new class of aggregation functions, ones that are particularly well suited for relational classification and class probability estimation. The empirical part of this paper demonstrates on real domain the effects on the system performance of different aggregation methods on different relational concepts. The results suggest that more complex aggregation methods can significantly increase generalization performance and that, in particular, task-specific aggregation can simplify relational prediction tasks into well-understood propositional learning problems.

A number of technology products display positive network effects, and are used in variable quantities by heterogeneous customers. Examples include operating systems, infrastructure and back-end software, web services and networking equipment. This paper studies optimal nonlinear pricing for such products, under incomplete information, and with the threat of competitive entry. Both homogeneous and heterogeneous network effects are modeled. Conditions under which a fulfilled-expectations contract exists and is unique are established. While network effects generally raise price, it is shown that accompanying changes in consumption depend on the nature of the network effects - in some cases, it is optimal for the monopolist to induce no changes in usage across customers, while in others cases, network effects raise the usage of all market participants. Optimal pricing is shown to include quantity discounts that increase with usage, and may also involve a nonlinear two-part tariff. These results highlight the impact of network effects on trade-offs between price discrimination and value creation, and have important managerial implications for pricing policy in technology markets. The need to deter competitive entry generally lowers profits for the monopolist, and increases customer surplus. When network effects are homogeneous across customers, the resulting entry-deterring monopoly contract is a fixed fee and results in the socially optimal outcome. However, when the magnitude of heterogeneous network effects is relatively high, there are no changes in total surplus induced by the entry threat, and the price changes merely cause a transfer of value from the seller to its customers. The presence of network effects, and of a credible entry threat, are also shown to increase distributional efficiency by reducing the disparity in relative value captured by different customer types. Regulatory and policy implications of these results are discussed.

We model an oligopolistic technology market in which firms endogenously choose product scope, fixed costs are affected by exogenous technological progress, and there may be threat of entry. Our analysis shows that equilibrium outcomes involve substantial overinvestment in product scope, which benefit consumers and hurt firms, relative to the social optimum. Technological progress generally increases consumer surplus and lowers firm profits. If entry is threatened bilaterally across two converging markets, both either accommodate entrants from the rival market, or both deter entry; continuous progress in technology can cause equilibria shifts, leading to discontinuous and radical redistribution of surplus across markets.

Thousands of business news stories (including press releases, earnings reports, general business news, etc.) are released each day. Recently, information technology advances have partially automated the processing of documents, reducing the amount of text that must be read. Current techniques (e.g., text classification and information extraction) for full-text analysis for the most part are limited to discovering information that can be found in single documents. Often, however, important information does not reside in a single document, but in the relationships between information distributed over multiple documents. This paper reports on an investigation into whether knowledge can be discovered automatically from relational data extracted from large corpora of business news stories. We use a combination of information extraction, network analysis, and statistical techniques. We show that relationally interlinked patterns distributed over multiple documents can indeed be extracted, and (specifically) that knowledge about companies’ interrelationships can be discovered. We evaluate the extracted relationships in several ways: we give a broad visualization of related companies, showing intuitive industry clusters; we use network analysis to ask who are the central players, and finally, we show that the extracted interrelationships can be used for important tasks, such as for classifying companies by industry membership.

A data mining (DM) process involves multiple stages. A simple, but typical, process might include preprocessing data, applying a data-mining algorithm, and postprocessing the mining results. There are many possible choices for each stage, and only some combinations are valid. Because of the large space and non-trivial interactions, both novices and data-mining specialists need assistance in composing and selecting DM processes. We present the concept of Intelligent Discovery Assistants (IDAs), which provide users with (i) systematic enumerations of valid DM processes, in order that important, potentially fruitful options are not overlooked, and (ii) effective rankings of these valid processes by different criteria, to facilitate the choice of DM processes to execute. We use a prototype to show that an IDA can indeed provide useful enumerations and effective rankings. We discuss how an IDA is an important tool for knowledge sharing among a team of data miners. Finally, we illustrate all the claims with a comprehensive demonstration using a more involved process and data from the 1998 KDDCUP competition.

This paper analyzes optimal pricing for information goods under incomplete information, when both unlimited-usage (fixed-fee) pricing and usage-based pricing are feasible. For a general set of customer characteristics, it is shown that in the presence of contract administration costs, offering fixed-fee pricing in addition to a non-linear usage-based pricing scheme is always profit-improving, and there may be markets in which a pure fixed-fee is optimal. Moreover, it is proved that the optimal usage-based pricing schedule is independent of the value of the fixed-fee. These results imply that the optimal pricing strategy is never fully revealing. A procedure for determining the optimal combination of fixed-fee and non-linear usage-based contracts is presented. Applying these general results to specific: business contexts suggests a number of operiG tional guidelines for designing pricing schedules, and managerial insights for setting pricing policy. For instance, in nascent information markets, firms are most likely to profit from low fixed-fee penetration pricing, but as these markets mature, the optimal pricing mix should expand to include a wider range of usage-based pricing options. The effects of changes in product value and administration costs on the adoption levels of different pricing schemes, optimal quantity discounts, firm profitability and total welfare are analyzed. Strategic pricing responses to changes in market characteristics are described, and the implications of the paper's results for bundling and vertical differentiation of information goods are discussed.

Tree induction and logistic regression are two standard, off-the-shelf methods for building models for classification. We present a large-scale experimental comparison of logistic regression and tree induction, assessing classification accuracy and the quality of rankings based on class-membership probabilities. We use a learning-curve analysis to examine the relationship of these measures to the size of the training set. The results of the study show several remarkable things. (I) Contrary to prior observations, logistic regression does not generally outperform tree induction. (2) More specifically, and not surprisingly, logistic regression is better for smaller training sets and tree induction for larger data sets. Importantly, this often holds for training sets drawn from the same domain (i.e., the learning curves cross), so conclusions about induction-algorithm superiority on a given domain must be based on an analysis of the learning curves. (3) Contrary to conventional wisdom, tree induction is effective at producing probability-based rankings, although apparently comparatively less so for a given training--set size than at making classifications. Finally, (4) the domains on which tree induction and logistic regression are ultimately preferable can be characterized surprisingly well by a simple measure of signal-to-noise ratio.

For many supervised learning tasks, the cost of acquiring training data is dominated by the cost of class labeling. In this work, we explore active learning for class probability estimation (CPE). Active learning acquires data incrementally, using the model learned so far to help identify especially useful additional data for labeling. We present a new method for active learning, BootstrapLV, which chooses new data based on the variance in probability estimates from bootstrap samples. We then show empirically that the method reduces the number of data items that must be labeled, across a wide variety of data sets. We also compare Bootstrap-LV with Uncertainty Sampling, an existing active-learning method for maximizing classification accuracy, and show not only that BootstrapLV dominates for CPE but also that it is quite competitive even for accuracy maximization.

Apart from reducing buyer search costs, web-based commerce has also enabled the use of intelligent agent technologies that reduce seller search costs by targeting buyers, customizing, and pricing products in real-time. Our model of an electronic market with customizable products analyzes the pricing, profitability and welfare implications of these agent-based technologies that price dynamically, based on product preference and demographic information revealed by consumers. We find that in making the trade-off between better prices and better customization, consumers invariably choose less-than-ideal products. Furthermore, this trade-off impacts buyers on the higher end of the market more, and causes a transfer of consumer surplus towards buyers with a lower willingness to pay. As buyers adjust their product choices in response to better demand agent technologies, sellers may experience reduced revenues, since the gains from better buyer information are countered by the lowering of the total value created from the transactions. We study the strategic and welfare implications of these findings, and discuss managerial and technology development guidelines.

The rapid growth and great popularity of Internet sites that specialize in providing intangible services in the form of information and community services gives rise to new forms of competition. Information Web sites such as www.how2.com and community sites such as www.iVillage.com provide free "content" and rely on advertising and hosting revenues to generate income. The competition between content sites in the same market niche is intense and only a few companies are likely to survive. In this paper, we examine a number of competitive models or "information games" that provide insights into the nature of this competition. The models capture differences between the type and maturity of the markets and differences in the behavioral assumptions about the nature of consumer demand for content. While these markets often have a "winner-takes-all" nature, we find a number of situations in which more than one player can survive at equilibrium.

It is difficult for the firm competing through information technology (IT) resources to gain a sustainable advantage because systems are easy to imitate and often substitute resources are available to competitors. The innovator may be unable to appropriate all of the benefits from IT investments. Airlines have installed computerized reservations systems (CRSs) in travel agencies in order to appropriate the returns from their investments in information technology. The airlines expected to obtain a number of benefits from this strategy including increased efficiency, possible bias in favor of the CRS owner on the part of the travel agent, and fees from other airlines for making reservations for them. The purpose of this paper is to evaluate the appropriation of value by CRS owners from deploying systems in travel agencies. These benefits, beyond fees from travel agents, should be seen in the vendor airline's market share between cities and in the overall performance of the airline at an industry level. This paper models airline performance as a function of CRS ownership at two levels: for selected city-pairs and at the overall level of the firm. The city-pair analysis employs a multinomial logit (MNL) market share model that analyzes five years of data on 72 city-pair routes. The industry model uses longitudinal data for a panel of ten airlines for twelve years. The results of both analyses support hypotheses that CRS ownership is positively related to airline performance. It appears that strong airlines have appropriated the benefits of their CRSs, turning them into highly specialized assets for further travel-related innovation. This work offers useful theoretical extensions and methodological approaches for the study of similar kinds of network technology innovations that are currently being deployed in association with electronic commerce on the Internet.

Organizations require mechanisms to efficiently distribute knowledge such as news releases, seminar announcements, and memos. While the machinery for information storage, manipulation. and retrieval exists, research dealing directly with knowledge distribution in an organizational context is scarce. In this paper, we address this need by first examining the pros and cons of the conventional "mailing lists" approach and then proposing new workflow mechanisms that improve the efficiency and effectiveness of knowledge distribution. The main contributions of this study include: (I) a workflow perspective on organizational knowledge distribution. (2) workflow analysis of two new knowledge distribution methods based on dynamic mailing lists and profile matching, respectively, and (3) a new way of matching knowledge supply and demand that extends existing information filtering algorithms.

We characterize the optimal dynamic price policy of a monopolist who faces "viscous" demand for its services. Demand is viscous if it adjusts relatively slowly to price changes. We show that with the optimal policy the monopolist stops short of achieving 100% market penetration, even when all of the consumers have the same long-run willingness to pay for the service. Furthermore, for certain parameter values in the model, the price policy requires rapid oscillations of the price path.

Profit centers in a firm in multidivisional form agree in the ex ante stage upon a plan about their joint production and profit imputation. The plan is executed in the subsequent two periods of the interim stage. In the first interim period, each center has its private information, but a part of its information is revealed to the other centers through its action. Based on the information endogenously pooled this way, the centers take another round of actions in the second interim period. A core plan is defined as a Bayesian incentive-compatible plan of the grand coalition of profit centers, upon which no coalition can improve using its Bayesian incentive-compatible plan. A core plan is called full-information revealing if each center fully reveals its private information in the first period. Three existence theorems for a full-information revealing core plan are established. The first two theorems impose alternative conditions on returns to scale: (1) the neoclassical convex technology, and (2) increasing returns to scale. In case (2), a stronger condition than Scarf's distributiveness is imposed on the total production set. The third theorem is based on a specific supplier-customer relationship among the divisions.

We use a model of real-time decentralized information processing to understand how constraints on human information processing affect the returns to scale of organizations. We identify three informational (dis)economies of scale: diversification of heterogeneous risks (positive), sharing of information and of costs (positive), and crowding out of recent information due to information processing delay (negative). Because decision rules are endogenous, delay does not inexorably lead to decreasing returns to scale. However, returns are more likely to be decreasing when computation constraints, rather than sampling costs, limit the information upon which decisions are conditioned. The results illustrate how information processing constraints together with the requirement of informational integration cause a breakdown of the replication arguments that have been used to establish nondecreasing technological returns to scale.

Information Systems Working Papers Series

As the business environment become more uncertain and competitive, many organizations are seeking ways to gain economic efficiency and share in business and technology risk. Despite wide differences in the reference disciplines applied to outsourcing research, the vast majority of it assumes a one-to-one relationship between the client and the outsourcing vendor. This paper examines the economic, strategic, and organizational issues involved in IT outsourcing when more complex arrangements are considered - such as multi-vendor alliances, co-sourcing, and complex multi-vendor, multi-client relationships. In this paper, we identify a taxonomy of four classes of outsourcing relationships (based on how many clients and vendors are involved in the outsourcing relationship, and illustrate each with recent business examples. Grounded in this taxonomy, we develop a theoretical framework that identifies both enabling and constraining forces that may influence client firms in choosing among the four types of outsourcing relationships. This paper provides insights regarding how the variations in the nature of these outsourcing relationships may shape the benefits and risks be achieved from outsourcing, as well as the ongoing complexity of managing outsourcing relationships.

Information Systems Working Papers Series

The Relationship Management Methodology (RMM) for hypermedia design was originally introduced in 1995, and has since evolved in a number of ways in response to the rapid growth in demand for hypermedia applications on the World Wide Web. The revamped methodology is demonstrated in the design of rich web applications. Design and implementation issues are discussed, including database integration, and top-down versus bottom up approaches to Web Information System (WIS) application development. The graphical and programming language notations for RMM's new constructs are presented. RMM promotes sound design and maintainable development of hypermedia.

This paper describes an Electronic Butler (or e-Butler) that provides a customer-centric personalized shopping services to its subscribers across a wide range of products. This service is provided by identifying individual customer's shopping needs from the comprehensive purchasing history of that person and providing purchasing recommendations or direct purchasing decisions for the customer. e- Butler service consists of two components -- the Personal Shopping Assistant (PSA) service that provides purchasing recommendations to the customer and the Magic Wand (MW) service that directly makes purchases it believes the customer needs without any prior consultations with the customer. In order to understand how PSA and MW services of e-Butler are related to the existing one-to-one marketing and recommender systems, a general framework classifying various personalized shopping services is presented that clearly delineates PSA and MW services from these existing systems. Moreover, the paper presents an architecture of the e-Butler service, explains what its business value is, discusses its feasibility, and describes what needs to be done to make it a successful service.

The Internet and the Web Information Systems (WIS) create new channels for communications between governments, corporations and individuals. Governments collect, generate and disseminate vast amounts of information. In a market driven and democratic society, this information is vital to enhance the trust of citizens in their government and institutions, and critical to individual and organizational decision-making. This paper describes lessons learned from the "EDGAR on the Internet" (EOI) project, an early demonstration web information system (WIS) for disseminating corporate disclosure documents filed with the U.S. Securities and Exchange Commission (SEC). Building on our experience with this system, and the emerging impact of the Internet on the market for disclosure documents, we provide guidelines for governments and information vendors to effectively adopt and adapt to web-enabled innovations for data dissemination.

We discuss the effects of strategic commitments and of network size in the process of setting interconnection fees across competing networks. We also discuss the importance of the principles of reciprocity and imputation of interconnection charges on market equilibria. Reciprocity means that both networks charge the same for interconnection. Imputation means that a network charges its customers as much as it charges customers of the other network for the same service. Assuming that each consumer cannot subscribe to more than one network, we begin by analyzing a game of strategic symmetry where the two networks choose all prices simultaneously. Second, we allow a dominant network to set the interconnection fee before the opponent network can set its prices. This results in a price-squeeze on the rival network. Third, we show that the imposition of a reciprocity rule eliminates the strategic power of the first mover. Under reciprocity, one network sets the common interconnection fee at cost, and the equilibrium prices for final services are lower than in the two previous games without reciprocity. Moreover, prices under reciprocity obey the principle of imputation. In the long run, consumers subscribe to one of the two networks. Typically, there is a multiplicity of equilibria, including corner equilibria, where all consumers subscribe to the same network. However, under reciprocity, there are no corner equilibria.

We show that application of the so-called "Market Determined Efficient Component Pricing Rule," the "Efficient Component Pricing Rule," and, in general, of pricing rules that are based on private opportunity costs would perpetuate pricing inefficiencies and result in lower social surplus than pricing which is based on social opportunity cost rather than private opportunity costs.

This paper studies the dimensions underlying user involvement with Websites and builds upon the existing body of knowledge on involvement with traditional media. A multidimensional bipolar semantic differential scale based on Zaichkowsky's Personal Involvement Inventory is used to identify the factors that determine the level of involvement in a Website. Websites are then classified, based on these factors, as high or low involvement sites. Involvement with Web sites was found to comprise of three dimensions - cognitive, affective and structural. Among these, the cognitive and the affective dimensions were found to have the highest discriminating power between high and low involvement sites while the structural dimension was found to serve as a moderating factor.

This paper reviews the current conditions in the U.S. telecommunications industry. It first examines the impact of technological and regulatory change on market structure and business strategy, and then the impact on pricing of digitization and the emergence of internet telephony. Then the paper considers the effects of the 1996 Telecommunications Act on market structure and strategy in conjunction with the history of regulation and antitrust intervention in the telecommunications sector. After discussing the impact of wireless technologies, the paper concludes with some short-term predictions, as well as concern about the derailment of the implementation of the 1996 Act by aggressive legal tactics of entrenched monopolists (the local exchange carriers), and the real danger that the intent of Congress in passing the 1996 Act to promote competition in telecommunications will not be realized.

Information Systems Working Papers Series

Manufacturing flexibility is critical for survival in industries characterized by rapid change and diverse product markets. Although new manufacturing technologies make it possible to accomplish flexibility, their potential remains unrealized by firms whose organizational elements do not possess adaptive capabilities. We use the brain as a metaphor to generate insights on how firms might design flexible production systems. We chose the brain as a metaphor because it is a self-organizing system capable of responding rapidly to a broad range of external stimuli. The brain as a metaphor suggests that flexibility can be enhanced by employing practices that promote distributed processes occurring in parallel manner. Such practices lie in contrast to those employed by production systems built on scientific management principles that promote localized processes in a sequential manner. By exploring these contrasting modes of operation, we argue that the brain as a metaphor opens up new avenues for theory development related to the design of flexible production systems.

Several key technologies are converging to create the emerging cyberspace. We characterize this convergence process as one of cumulative synthesis and suggest that the network mode of organization is the most appropriate for facilitating convergence.

Technological systems are shaped both by forces arising from the technical environment of product markets and those arising from the institutional environment of compatibility standards. We explore how it might be possible for standards to simultaneously enable activities in the technical environment and not constrain them. Such a scenario is possible when the technical environment is not completely embedded in the standards that shape them. We characterize such technological systems as being "just" embedded.

Both researchers and managers depend on the accuracy of managers' perceptions. Yet, few studies compare subjective with "objective" data, perhaps because it is very difficult to do well. These difficulties also muddy interpretations of results. On one hand, studies suggest that managers' perceptions may be very inaccurate. On the other hand, the observed errors in managerial perceptions may arise from research methods instead of managers. Because perceptual data are so significant for both researchers and managers, researchers need to understand both the potential contaminants of perceptual research and the determinants of perceptual errors and biases. This article reviews studies of the accuracies of managers' perceptions, points out hazards in such research, and suggests various ways to improve studies of perceptions. The suggestions encompass improvements in gathering more valid subjective data, locating more appropriate "objective" data, finding appropriate respondents, and using statistical methods that provide accurate and reliable estimates with small samples.

Optimal arrangements of cartridges and file partitioning schemes are examined in carousel type mass storage systems using Markov decision theory. It is shown that the Organ-Pipe Arrangement is optimal under different storage configurations for both the anticipatory as well as the non-anticipatory versions of the problem. When requests arrive as per an arbitrary renewal process this arrangement is also shown to minimize the mean queueing delay and the time spent in the system by the requests.

In this paper, we describe the architecture of an object-oriented scheduling system. First, a mathematical framework is presented that is based on set theory and graph theory. Then a number of basic as well as more specialized methods are defined which can be applied on the entities of any decision support system. The principal objects of a scheduling system are defined, as well as the methods specifically designed for the manipulation of the schedules. The object base design, the schedule generator design and the user interface design are then discussed in detail.

A number of unstructured or partially structured electronic secondary markets exist to enable the sale and trading of goods between consumers. Many tend to be self-administering UseNet groups, or WWW sites for niche products; however, there has been significant recent growth in the number of more general web-based markets of this kind. Apart from facilitating reliable and liquid trade of used goods, the existence of these markets can alter the desirability of new product, as well as products that are complementary/compatible, and one expects to see a proliferation of such trading forums as Internet technology continues to become more widespread and reliable, and less expensive. We present a economic framework for analyzing how these electronic secondary markets affect the demand for a primary product. We then examine when it is optimal for a firm to operate a market of this kind, and when their presence is socially optimal. Surprisingly, we find that in a number of cases, the presence of these markets has a primary positive effect on the profitability of a new good; this leads us to conjecture that there will soon be a number of such trading forums operated by manufacturers of primary goods. We also find that in a majority of cases, it is feasible for a third-party intermediary to profitably operate such a market. Key parameters that affect the desirability of the market are the existing installed customer base, the cost of information technology, the durability of the products in question, their rate of technological obsolescence and the nature of customer preferences.

An Asynchronous Learning Network (ALN) is a Computer-Mediated Communication System designed to support "anytime/anywhere" interaction among students and between students and instructors. A field experiment compared groups and individuals solving an ethical case scenario, with and without an ALN, to determine the separate and joint effects of communication medium and teamwork. Undergraduate students in Computers and Society analyzed the case as an assignment in the course. Dependent variables include quality of the reports, learning as measured by similar cases on the final exam, and subjective perceptions of learning. The results indicate that working in a group, instead of alone, tends to increase motivation, perception of learning and solution satisfaction. Individuals working online produced higher quality reports on the ethics scenario than individuals working manually, and computer-supported groups produced the longest reports, while individuals working manually produced the shortest reports. Regarding group conditions, manual teams reported significantly higher levels of process satisfaction, perception of process structure and perception of discussion quality than teams supported by an asynchronous communication medium. However, computer-supported groups reported the highest levels of perceived learning. Finally, perception of collaborative learning does not seem to be affected by the use of the medium; both supported and unsupported groups perceived about the same levels of collaborative learning.

Information technology has radically altered the management of supply chain operations; many business partners who are adjacent on the supply chain can gain from entering inter-organizational information sharing (IOIS) relationships and sharing information that was previously accessible to only one of them. This situation is typical in retailer-supplier logistics management relationships. The first part of our study analyzes different forms of virtual integration - relationships between independent companies that result in some of their operations resembling those of a single vertically integrated firm - and classifies them based on their models of information sharing across the supply chain. We find that there are four primary policies that firms adopt when they exchange information across the supply chain; these are EDI, vendor managed inventory (VMI), continuous replenishment (CR) and category management (CM). Typically, corporations view the development of inter-organizational information systems, and the sharing of information as being targeted at increasing operational efficiency by reducing ordering costs, inventory costs and supply lead times. Many studies have focused on studying IOIS technology issues, and estimating the value generated from these arrangements using traditional models of inventory and ordering costs. However, we find that in a number of cases, the information shared can have cross-functional value - it can also be used to improve a supplier's production planning, and to alter their marketing and sales strategies. Paradoxically, however, suppliers who receive such information feel that not only are their benefits minimal, but they often end up worse off than before the IOIS was implemented. The second part of our study explains this paradox. We show how retailers and other buyers can successfully contract to end up with more value than is generated by the sharing of information. Using game-theoretic models of strategic interaction, we show that this effect intensifies as the competitive value of the information to the supplier's marketing and sales departments increases. Besides, as the value that could be generated by the sales and production divisions of the supplier increases, we demonstrate how the supplier loses more and more value. Furthermore, the buyer need not actually share the information to derive these rents; we indicate why the possibility of sharing is sufficient, even when the buyer cannot independently create value from that information. The practical contributions of this inter-disciplinary study are manifold. We provide a clear and lucid description of the different levels at which organizations share information. We also describe a fairly general modeling framework which lays the foundation for a deeper analysis of this increasingly important area. Our strategic results demonstrate that a single focus on the technological or operational aspects of IOIS can mislead managers significantly. The true costs and benefits of these relationships can only be judged by recognizing the cross-functional impact of the information flows on the operational architecture, the marketing strategies of the suppliers and buyers, and the nature of competition within the respective organizations' industries.

We describe a quantitative model for the performance evaluation of interactive computer systems. The approach involves the development of an "interaction graph" or state transition diagram to describe the user-machine interaction. Given numerical data on transition times and probabilities, the model can be used to perform sensitivity analyses of changes in system parameters and user behavior. To illustrate the model, we use empirical data from field and laboratory experiments designed to compare a prototype natural language query system with a formal (relational) query system. The general approach is applicable in a broad range of other contexts including bibliographic retrieval and the analysis of web-log data. It should be of interest to both system developers and potential users of these systems.

The introduction of computerized post-coordination has solved many of the problems of pre-coordinated subject access. However, the adoption of computerized post-coordination results in the loss of some precoordination benefits. Specifically, the effect of hiding terms within the context of others is lost in post-coordination which gives lead status to every document term, This results in spurious matches of terms out of context. Library patrons and Internet searchers are increasingly dissatisfied with subject access performance, in part because of unmanageably large retrieval sets. The need to enhance precision and limit the size of retrieval sets motivates this work which proposes partial coordination, an approach which incorporates the advantages of computer search with the ability of precoordination to limit spurious partial matches and thereby enhance precision.

In time series problems, noise can be divided into two categories: dynamic noise which drives the process, and observational noise which is added in the measurement process, but does not influence future values of the system. In this framework, empirical volatilities (the squared relative returns of prices) exhibit a significant amount of observational noise. To model and predict their time evolution adequately, we estimate state space models that explicitly include observational noise. We obtain relaxation times for shocks in the logarithm of volatility ranging from three weeks (for foreign exchange) to three to five months (for stock indices). In most cases, a two-dimensional hidden state is required to yield residuals that are consistent with white noise. We compare these results with ordinary autoregressive models (without a hidden state) and find that autoregressive models underestimate the relaxation times by about two orders of magnitude due to their ignoring the distinction between observational and dynamic noise. This new interpretation of the dynamics of volatility in terms of relaxators in a state space model carries over to stochastic volatility models and to GARCH models, and is useful for several problems in finance, including risk management and the pricing of derivative securities.

We develop a model to predict 1) the use of a multifunctional, broker workstation with a windowed interface and 2) the relationship between workstation use and performance. Brokers and sales assistants in the private client group of a major investment bank use this workstation as an integral part of their jobs. Our model explains some of the variance in their usage, intended usage and performance, but the variables that are most salient in the model differ between brokers and sales assistants. There is evidence that low performing brokers use the workstation more than higher performing brokers; the results also suggest that a different type of training may be needed for sophisticated workstations for professionals than for clerical personnel learning to use transactions processing systems. We believe it is important to understand the acceptance of technology and the relationship between system use and performance if firms are to obtain a return from investing in information technology.

Organizations are taking advantage of "data-mining" techniques to leverage the vast amounts of data captured as they process routine transactions. Data-mining is the process of discovering hidden structure or patterns in data. However several of the pattern discovery methods in datamining systems have the drawbacks that they discover too many obvious or irrelevant patterns and that they do not leverage to a full extent valuable prior domain knowledge that managers have. This research addresses these drawbacks by developing ways to generate interesting patterns by incorporating managers' prior knowledge in the process of searching for patterns in data. Specifically we focus on providing methods that generate unexpected patterns with respect to managerial intuition by eliciting managers' beliefs about the domain and using these beliefs to seed the search for unexpected patterns in data. Our approach should lead to the development of decision support systems that provide managers with more relevant patterns from data and aid in effective decision making.

Virtual organizations are new organizational forms comprising a set of network transactions that differ from those found in markets and hierarchies. This paper explores the nature of these network transactions through an in-depth study of a virtual firm. The virtual organization is characterized by constant organizing through virtual teams and alliances, a unique management culture and a set of norms, information and knowledge sharing enabled by information technology, and employee self-governance. The organization gains from a culture of fast-response and efficiency while employees are trusted to exercise discretion and take initiatives.

This paper describes a new approach to measuring the complexity of software systems with considering their readability. Readability Metrics were first proposed by Chung and Yung 181 in 1990. Software industry uses software metrics to measure the complexity of software systems for software cost estimation, software development control, software assurance, software testing, and software maintenance [3], [71, [9], 151, [18]. Most of the software metrics measure the software complexity by one or more of the software attributes. We usually class@ the software attributes that software metrics use for measuring complexity into three categories: size, control flow, and data flow [5], f71. All the three categories concern with the physical activities of software development. Readability Metrics have been outstanding among the existing software complexity metrics for taking nonphysical software attributes, like readability, into considerations [8]. The applications of Readability Metrics are good in indicating the additional efforts required for less readable software systems, and help in keeping the software systems maintainable. However, the numerous metrics and the complicated formulas in the family usually make it tedious to apply Readability Metrics to large scale software systems. In this paper, we propose a simplified approach to Readability Metrics. We reduce the number of required measures and keep the considerations on software readability. We introduce our Readability model in a more formal way. The Readability Metrics preprocesses algorithm is developed with compilers front-end techniques. The experiment results show that this simplified approach has good predictive power in measuring software complexity with software readability, in addition to its ease of applying. The applications of Readability Metrics indicate the readability of software systems and help in keeping the source code readable and maintainable.

The Internet and similar networks provide new infrastructures for communications and commerce. These open networks interconnect computers across many different organizations with dramatically lower communications and distributed applications development costs. This motivates businesses to transfer commercial activity from closed private networks to open networks like the Internet. However, open network architectures are vulnerable to a number of different security threats. While many different hardware and software solutions exist to secure transactions over the Internet, greater consensus is required by companies and consumers on the processes, organizations and application of existing technical solutions for secure electronic commerce. Greater consensus on security among trading parties will lower the costs of electronic commerce and accelerate its deployment on the Internet.

Advances in information technologies enable firms to expand electronic commerce - the exchange of valuable information, goods and services across electronic media. This paper looks at how businesses are adopting and using technologies like the Internet and World Wide Web for electronic commerce. I propose that electronic commerce will dramatically reduce transactions costs requiring managers to re-develop their firm's strategy to focus on more fundamental sources of competitive advantage. I identify responses required to compete in a low transactions cost environment and provide illustrative examples of firms executing such a strategy.

While many managers face the challenge of lower profits in increasingly competitive and commoditized industries, a few firms break out as market value leaders generating superior growth and shareholder returns. How do these firms break out of the commoditization trap? In this paper we propose these firms invent unique value propositions and offer them in superior ways to their customers. Based on our study of a number of market leaders, we provide a framework for systematically understanding and re-inventing the firm's value propositions. We propose that clearly understanding and defining the dimensions of a firm's value proposition is a critical first step in building an effective strategy.

It is difficult for the firm investing in information technology (IT) to appropriate a1 of the benefits from its investment for itself- it is very easy to imitate innovations in IT. Airlines have installed computerized reservations systems (CRSs) in travel agencies in order to appropriate the returns from their investments in information technology. The airlines expected to obtain a number of benefits from this strategy including increased efficiency, possible bias in favor of the CRS owner on the part of the travel agent, and fees from other airlines for making reservations for them. The purpose of this paper is to evaluate the impact of the indirect (non-fee) benefits to CRS owners from deploying systems in travel agencies. These indirect benefits should be seen in the vendor airline's market share between cities and in the overall performance of the airline at an industry level. This paper models airline performance as a function of CRS ownership at two levels: for selected city-pairs and at the overall level of the firm. The city-pair analysis employs a multinomial logit market share model using five years of data on 72 routes. The industry model uses longitudinal data for a panel of ten airlines for twelve years. The results of both analyses support hypotheses that CRS ownership is positively related to airline performance, It appears that strong airlines have appropriated the indirect benefits of their CRSs, turning them into highly specialized assets for further travel-related innovation.

This paper presents a field study of the Technology Acceptance Model. We extended this model to predict the acceptance of a multifinctional, broker workstation with a windowed interface. Brokers and sales assistants in the private client group of a major investment bank use this workstation as an integral part of their jobs. The extended model explains a significant percentage of the variance in usage, but the variables that are most salient in the model differ between brokers and sales assistants. There is evidence that low performing brokers use the workstation more than higher performing brokers; the results also suggest that more training may be needed for sophisticated workstations for professionals than for clerical personnel learning to use transactions processing systems. We believe it is important to predict and understand the acceptance of technology like the workstation in this study if firms are to obtain a return from investing in information technology.

We use computer-based simulations of a stock market as a background environment for experimental tests of the integration of an order-driven trading system into a dealer/quote-driven market. Experimental subjects traded using a traditional dealer quote screen (such as Nasdaq in the U.S. or the London Stock Exchange's SEAQ), to which was added a public limit order facility. Data captured on subjects' trading decisions under different market structures revealed that: (1) When available, the limit order facility was used by the subjects, attracting some orders that would have otherwise gone to dealers, and reducing investor trading costs. (2) The relative use of market orders and limit orders was related to the bid-ask spread; wider spreads (higher cost of immediate trading) led subjects to enter fewer market orders. (3) Limit order use was reduced when the dealers were provided with an "informational advantage. " (4) While the introduction of a limit order facility did not have a substantial effect on dealer profit margins, dealers' activities as a percentage of total market volume declined. Overall, we find the simulation environment is a workable device for analyzing the effect of market design changes on trader behavior and market quality. It can provide solid guidance on market structure issues, such as how best to incorporate a limit order facility in a competing dealer market.

This briefing focuses on design and development of WWW systems. I will present the principal elements of the Relationship Management Methodology which aids in the design and development of WWW applications. RMM will be introduced via a sample application. I will also, simultaneously, demonstrate a software tool, RM-CASE, that provides computerized support got RMM.

Finns invest in information technology (IT) to create various kinds of leverage on firm profitability and performance. However, IT researchers have concentrated their efforts on the productivity impacts of technology, at the employee, process, firm, industry, and economy levels of analysis, to the exclusion of other business value impacts. Not captured by productivity metrics are the significant benefits that may accrue to the firm as product quality improves, managerial assessment of risk is enhanced, time to market and other cycle time reductions are made, and new ways to control firm input and output prices become available to management. These kinds of impacts reflect price recovery improvements - the ratio of the prices of a firm's outputs (of goods and services) to the prices of the inputs it consumes in production - and they are rarely measured or understood in a systematic way. In this paper, we argue that it is appropriate to reconsider the current measurement and research agenda that aims to discover and document the payoffs that accrue from corporate investments in IT. We illustrate the extent to which IT investment may be motivated by management's understanding of the potential price recovery payoffs (even if they fail to carefully measure or report them) in the context of trading and treasury operations in international banking. We find that price recovery captures a previously unmeasured dimension of the business value of IT.

We study the properties of real-time decentralized information processing, as a model of human information processing in organizations, and use the model to understand how constraints on human information processing affect the returns to scale of firms. With real-time processing, decentralization does not unambiguously reduce delay, because processing a subordinate's report precludes processing current data. Because decision rules are endogenous, delay does not inexorably lead to eventually decreasing returns to scale; however, returns are more likely to be decreasing when computation constraints, rat her than sampling costs, limit the information upon which decisions are conditioned. The results illustrate that the requirement of informational integration causes a breakdown of the replication arguments that are often used to establish non-decreasing returns.

While "now" is expressed in SQL as CURRENT-TIMESTAMP within queries, this value cannot be stored in the database. However, this notion of an ever-increasing current-time value has been reflected in some temporal data models by inclusion of database-resident variables, such as "now," "until-changed," "∞," "@" and "-." Time variables are very desirable, but their use also leads to a new type of database, consisting of tuples with variables, termed a variable database. This paper proposes a framework for defining the semantics of the variable databases of temporal relational data models. A framework is presented because several reasonable meanings may be given to databases that use some of the specific temporal variables that have appeared in the literature. Using the framework, the paper defines a useful semantics for such databases. Because situations occur where the existing time variables are inadequate, two new types of modeling entities that address these shortcomings, timestamps which we call now-relative and now-relative indeterminate, are introduced and defined within the framework. Moreover, the paper provides a foundation, using algebraic bind operators, for the querying of variable databases via existing query languages. This transition to variable databases presented here requires minimal change to the query processor. Finally, to underline the practical feasibility of variable databases, we show that database variables can be precisely specified and efficiently implemented in conventional query languages, such as SQL, and in temporal query languages, such as TSQL2.

The introduction of computerized post-coordination has solved many of the problems of pre-coordinated subject access. However, the adoption of computerized post-coordination results in the loss of some pre-coordination benefits. Specifically, the effect of hiding terms within the context of others is lost in post-coordination which gives lead status to every document term. This results in spurious matches of terms out of context. Library patrons and Internet searchers are increasingly dissatisfied with subject access performance, in part because of unmanageably large retrieval sets. The need to enhance precision and limit the size of retrieval sets motivates this work which proposes partial coordination, an approach which incorporates the advantages of computer search with the ability of pre-coordination to limit spurious partial matches and thereby enhance precision.

This paper examines how firms use the Internet and the emerging "global infostructure" to improve key processes and competitive capabilities. It proposes the widespread diffusion of the emerging "infostructure" will dramatically reduce transaction costs, leading to growth in electronic commerce and productivity while reducing the profit opportunities of inefficient firms. In this wired world characterized by low transaction costs and noisy environments, firms will have to re-focus their strategies to achieve market leadership and capture mind share for competitive advantage. To achieve market leadership firms must seek advantage from economies of scale, scope, or externalities. Alternatively they must differentiate themselves through unique merchandise and innovation or develop specialized customer relations. To capture mind share firms must develop integrated cross media marketing programs. These programs will push products through various online or related promotions or develop strong brand identities for products. Thus this paper identifies competitive issues in a wired world and outlines various foci for strategic responses to the above challenges.

This paper presents a measurement instrument for diagnosing the level of data integration at the organization and organization subunit levels. Incorporating consideration for specific theoretical properties related to data integration into its design, the measurement instrument relies on a discrepancy (or gap) measure to adequately account for these properties. Data from forty-eight organizations of the Group Insurance industry are used to test the reliability and validity of the measurement instrument. The results suggest that the discrepancy measure, based on the difference between respondents' ideal (i.e. normative) and actual estimates of data integration, appears to be a valid indicator of data integration. However the efficacy of using a discrepancy measure, over the simpler use of its parts, must be considered in light of practical and theoretical considerations.

The Internet and the emerging global infostructure pose new opportunities and threats to businesses. This paper looks at the implications of these emerging low cost communications infrastructures on business practice and strategy. I examine how firms are using the Internet to improve innovation, production, sales, and service processes and I consider the effects of lower communications and coordination costs on business practices and strategies. Specifically, I propose that widespread use of the Internet and related technologies will dramatically reduce transaction costs, leading to a growth in electronic commerce and productivity. At the same time, the reduction in transactions and coordination costs will reduce the profit opportunities of inefficient firms, requiring them to re-focus their strategies. This paper identifies foci for strategic responses to the above challenges.

The internationalization of financial markets and the increasing demand for risk management products have fueled the growth of derivatives markets. While most exchanges have experienced increasing volumes over recent years, the pace of growth varies widely across exchanges, and the established marketplaces face increasing competitive pressures. In this paper, we investigate whether the trading mechanism offered to derivatives investors influences growth in market volumes. In particular, we distinguish between manual open outcry and electronic trading. In a floor market, traders gather in a pit and announce their orders. They complete trades using a combination of hand signals and eye contact. In an electronic market, orders are submitted to a central order book, and trades are created according to a matching algorithm. Using volume data from 1990-1994 for futures and options exchanges worldwide, we compute growth rates for the largest contracts and find that contracts traded in screen-based exchanges have experienced faster growth than those traded in manual markets. We discuss several interpretations of the data, but conclude that electronic exchanges are developing a competitive advantage.

Information Systems Working Papers Series

Interorganizational information systems (IOS) are increasingly relied on to facilitate the electronic exchange of data among organizations. Though substantive progress by the research community toward understanding IOS technology has been made, one unattended issue is understanding more fully the efficacy of alternative management strategies for assimilating IOS into organizational processes. Referred to here as IOS assimilation strategies, these strategies may offer to management useful alternatives for planning for and proceeding with IOS implementation. Conceptually based on Keen's (1991) notions of "reach" and "range" regarding organizations' computing infrastructure, two IOS Assimilation Strategies are tested for their comparative effects on three efficiency and six effectiveness measures. The empirical results, based on data from forty-eight organizations of the Group Insurance industry and the nonparametric Friedman test for significance testing, indicate that a significantly different ranking on the efficiency measures occurs across groups inhering varying strengths of IOS use pursuant to the alternative IOS Assimilation Strategies. The results suggest that greater efficiency gains may obtain by pursuing a strategy to extend IOS range over IOS reach. Consequently, management is advised to weigh carefully which IOS benefits, in terms of efficiency and effectiveness, are more important for substantiating IOS investments, and to make decisions regarding IOS implementation accordingly.

Much has been written about transforming organizations. At the heart of this activity is the notion of Business Process Redesign (BPR), the restructuring of a firm's basic business methods. We describe some of the forces that are behind this need for major organizational change. Next, we provide an overview of BPR and raise several questions concerning its suitability as a theoretical underpinning for radical organizational transformation. We then describe the transformation of Oticon, an international manufacturing firm with headquarters in Denmark. We have called this transformation a Metamorphosis because of the complete reconceptualization of the firm that took place. Using the example of Oticon, we speculate about the prerequisites for change of this extent, paying particular attention to the role of technology, which has been enabling rather than driving.

In many organizations, both business and scientific, we collect ever increasing amounts of data using information technology. Indeed, the technology for collecting data has outpaced our ability to analyze and interpret these very large databases. In this paper, we discuss the interaction of heuristic search and domain knowledge in the AX knowledge discovery tool. The search process rests on the use of rule quality measures and the organization of domain knowledge. A small loan application database from the machine learning repository is used to illustrate the process.

Very Large Scale Systems (VLSS) play a powerful role in shaping what an organization does and can do in a practical sense. VLSS are deeply embedded in the organizational procedures, business plans, and strategies. These systems evolve over long periods of time, often not according to some rational plan, and for a limited time they provide a significant competitive advantage over other firms. In the long run, however, VLSS become strategic liabilities and must be rebuilt. Many organizations experience great difficulty rebuilding VLSS . Indeed, most organizations attempt to avoid rebuilding VLSS until the last possible moment. Often, the organization is in a state of crisis, a strategic transition. Because of the complexity and size of VLSS, existing methodologies often are not helpful. To make matters worse, the typical management incentive structure discourages rebuilding VLSS. In a typical VLSS effort, participants soon discover that they must rebuild the organization in order to take full advantage of new technologies. A major organizational engineering effort is often required. Senior management as well as systems management routinely underestimate the complexity of the task before them. Consequently, large errors are made in estimating costs and time. Drawing on research in both the private and public sector, this paper examines why VLSS fail, why are VLSS so difficult to rebuild, what are the strategy options, and how can senior management guide the rebuilding process.

Previous literature on IT and productivity does not take into account different organizational goals and different management strategies for achieving these goals. But productivity and ROI relationships can easily differ as organizational goals and management strategies differ. Therefore, we argue, it is no longer appropriate to ask, "Does IT lead to productivity enhancement." or "Is the ROI on IT investments large or small or nonexistent? The better question is under what conditions of organizational climate and management choice does IT enhanced productivity result. To illustrate the powerful effect of organizational goals and management strategy on IT-productivity relationships, we examine the twenty year history of two of the largest IT users in the world: the Internal Revenue Service and the Social Security Administration. And we find that these two very similar agencies experienced very different results from massive investments in IT despite sharing a similar production function. There is nothing in micro economics however to explain the different strategies pursed by these managers. Instead we must turn to political and sociological models of organizations to understand the social construction of productivity results.

Interorganizational Systems (10s) are increasingly used to support exchange of business documents in many industries. Though clear improvements in organization performance are forthcoming in many cases, the research evidence is based largely on anecdotes and suggests that performance improvements may not be uniformly attained across organizations. Though integration of IOS with internal systems is recommended, the affects afforded through integration remain largely unexplored. This research study formulates a theoretical model of the complex interplay of relationships among IOS Usage, Systems Integration and Organization Performance, and tests the model's validity using a dataset of 48 organizations in the Group Insurance industry. The mode1 recognizes two conceptually distinct ways in which IOS may be used more intensively, includes two notions of systems integration which are argued to differentially affect the organization performance advantages, and uses a multidimensional organization performance construct to more adequately reflect the diversity of anticipated organization performance impacts presumed to extend from IOS use. The results are mixed in terms of direct organization performance impacts, and suggest that both notions of integration play a key role in facilitating attainment of the organization performance advantages.

We present the design of a computer-aided environment, RMCase, to support the design and construction of hypermedia applications. The environment is based upon the Relationship Management methodology. MCase supports hypermedia design and development activities. Support for cognitive design processes is achieved through three fundamental premises that form the foundation of RMCase: (1) fluid feedback loops between the various methodological stages, (2) manipulation of objects at the instance level, and (3) lightweight prototyping . To achieve this, RMCase itself is designed as a hypermedia application, where hypertextual navigation implements feedback loops. Instance objects can be cloned and abstraction/instantiation mechanisms are envisioned to facilitate designers back and forth movements between the abstract and the concrete layers of an application. As a result, RMCase will support bottom-up, top-down and middle-out software development styles.

The Dutch flower auctions have played a critical role in the world cut-flower industry by providing efficient centers for price determination and transactions of flowers between buyers and sellers. These auctions owned by Dutch cut-flower grower cooperatives have traditionally used the “Dutch auction” as the mechanism for price determination. This paper considers how changing patters of international competition, buyer preferences and information technology are likely to effect the organization of the Dutch flower auction. We provide a framework for analyzing the merits of different transaction models and use this framework to evaluate the strengths and weaknesses of existing and proposed electronic auction models for trading flowers. We propose information technology will enable new forms of trading that will partly replace and supplement the traditional Dutch auction as a method of organizing price determination and transactions. We identify how electronic trading will differ from prior mechanisms, and consider key challenges to the implementation of new auction models. Specifically we illustrate how the current auctions have been structured to serve the interests of growers, while electronic markets will primarily benefit buyers. Thus we highlight the importance of altering incentive and ownership structures in the Dutch flower industry to effectively transition to new electronic markets. This case illustrates the various complex issues that arise in the design and implementation of electronic markets, in settings characterized by changing technologies, pre-existing organizational processes and power structures.

Hypermedia design, as any other design activity, may be observed according to two points of view: methods which suggest milestones to guide the designer's work and process which concerns the actual detailed behavior of the designer at work. Cognitive studies assess that mental processes involved in any design process show widely shared human characteristics regardless to the used design method. Thereby, they provide general keys to help designers. Thus, a hypertext design environment should equally consider the two dimensions of a hypertext design activity, in particular it should support the natural design process specificities, mainly the incremental and opportunist aspects. The paper focuses on the hypertext design as a computer supported human activity. It examines what is general both in the design methods and in the design process of hypertexts in order to determine which general features are helpful to designers. This analysis has raised from the observation of the behavior of MacWeb users during design tasks. It is related to sound and well known results in cognitive science. The paper also describes how the proposed features are implemented in the MacWeb system.

Very-Large Scale Systems (VLSS) play a powerful role in shaping what an organization does and can do in a practical sense. VLSS are deeply embedded in an organization's procedures, business plans, and strategies. These systems evolve over long periods of time, often not according to some rational plan, and for a limited time they provide a significant competitive advantage over other firms. In the long run, however, VLSS become strategic liabilities and must be rebuilt. Many organizations experience great difficulty rebuilding VLSS. Indeed, most organizations attempt to avoid rebuilding VLSS until the last possible moment. Often, the organization is in a state of crisis, a strategic transition. Because of the complexity and size of VLSS, existing methodologies often are not helpful. To make matters worse, the typical management incentive structure discourages rebuilding VLSS. In a typical VLSS effort, participants soon discover that they must rebuild the organization in order to take full advantage of new technologies. A major organizational engineering effort is often required. Senior management as well as systems management routinely underestimate the complexity of the task before them. Consequently, large errors are made in estimating costs and time. Drawing on research in both the private and public sector, this paper examines why VLSS fail, why are VLSS so difficult to rebuild, what are the strategy options, and how can senior management guide the rebuilding process.

Outsourcing of services is usually viewed as a way to increase or decrease capacity at the margin. By outsourcing a firm can vary its ability to provide goods or services such as information technology (IT) processing capacity or applications development, without taking on the long term commitment of permanent staff or equipment Today, outsourcing provides new options for managers to provide information technology infrastructures and services to a firm. In this paper we identify ways in which outsourcing may be used tactically to better align business and information technology strategies. Specifically, we consider how selective outsourcing enables alignment of incentives, business revenues and IT infrastructure costs, and how organizational intent can be made more consistent with a firm' s capabilities. We propose that outsourcing as an information systems management strategy can enable better control and lower agency costs as managers benefit from better information through external benchmarking. We also consider the role of outsourcing as a vehicle to support major organizational transformation, and as a device for enabling new models of organization. Based on our analysis we propose a series of testable propositions about outsourcing. The contributions of this paper include extension of business strategy and information technology alignment concepts to incorporate the interdependence between projects, product life cycles, and technology transfer. Most importantly, the paper highlights new directions for outsourcing research, focusing on the effects of outsourcing on the IT organization, the processes of internal technology development and the transfer of technologies between organizations.

This paper reports the puzzling results of a study which examined IT capital investment and productivity at three of the largest IT user sites in the U.S. for the period 1970-1990: Social Security Administration (SSA), Internal Revenue Service (IRS), and the Federal Bureau of Investigation (FBI). Based on detailed IT investment, employment, and output data over twenty years, we found that only one agency had achieved significant productivity benefits, a second agency had modest results, and a third agency achieved no results whatever. These results cannot be explained by traditional theories of productivity of how productivity is produced. We argue that IT-induced productivity results not simply from strategic choice, nor the operation of the invisible hand in the market place, nor simply from keen managers adjusting their organizations to an "objective" environment. Instead we propose instead a new theory in which productivity benefits derive from a larger macro-culture enacted by powerful institutions in an organizational field. We extend this analysis to the larger economy and examine how this new theory helps us understand recent claims that IT is finally having positive productivity benefits at the sector level, and also helps us understand how the current fascination with reengineering and downsizing may be a self-fulfilling prophecy.

This paper argues that information technology can have a significant impact on organizational flexibility. Information technology (IT) contributes to flexibility by 1) changing the nature of organization boundaries and the time when work occurs 2) altering the nature and pace of work and 3) helping firms respond to changing market conditions. But, there are also aspects of technology which can decrease flexibility, and there may be second-order impacts of flexibility that are not easily predicted. Examples to illustrate the impact of information technology on two industries and three companies are presented. The paper concludes that management should consider the use of information technology to increase flexibility and suggests strategies for implementing flexible systems.

Software reuse in the presence of a repository and object-based CASE tool is likely to be "biased" Prior research has shown that a developer will be: (1) most likely to reuse her own objects; (2) somewhat less likely to reuse objects developed by her project team members; and, (3) even less likely to reuse objects stored in the repository, but developed elsewhere in the corporation. These biases can result in sub-optimal levels of software reuse. In the presence of such biases it is appropriate to deploy tools that support the search for software reuse, so that developers find it easier to reuse software objects authored by developers other than themselves or project team members. However; the tools that are chosen or created for this purpose must adequately treat the technical and cognitive fundamentals of the problem for individual developers, and recognize the organizational and economic perspectives of a firm that wishes to maximize the business value of its software development activities. In this paper we present a two-stage descriptive model that represents the search process for reusable software objects. We evaluate appropriate technologies, propose a technical solution to the problem of searching for reusable objects, and demonstrate its feasibility via a prototype implementation. The technical tool combines an automated classifier and a hypertext system. We describe an architecture to automatically create hypertext networks based on the classification schema. We illustrate our architecture using a classification of software objects obtained through structured interviews with software developers.

The increased use of network communications within industries, and among firms, suppliers, and customers, is focusing greater attention on the methods and standards for interorganizational communications. In the securities indushy, the settlement and clearing of trades depends on numerous messages to be sent and received by several organizations. Using the principles of Formal Languages for Business Communication (FLBC), we develop a message representation that is flexible and self-describing, and show how defeasible reasoning applied to settlement messages could handle problem trades. This application of FLBC offers advantages through machine-to-machine error reconcilement, integration with other market communications systems, and robustness to changes in securities design and regulation.

When decision makers want to know outcomes of business processes in their organizations, they often use simulations to do this. This paper describes how a new Query-Driven Simulation (QDS) approach can be used by decision makers to obtain information about future outcomes of business processes in a more declarative, flexible, and interactive way than the traditional approach of running simulations and then gathering statistics about simulation outcomes. The paper also describes the types of questions decision makers ask about outcomes of business processes and studies how easy it is to express these questions in terms of an SQL-like query language SimQL designed for Query-Driven Simulations. It also identifies the types of applications that are especially well-suited for QDS. Finally, the paper describes the Query-Driven Simulation Modeling Lifecycle and how QDS provides a feedback loop in the model development process.

We describe a step-by-step methodology for the design and construction of hypermedia applications and illustrate our approach using a small application. The Relationship Management Design (RMD) methodology begins with a data model of the application domain and proceeds through the design of the hypertext network, user interface and run-time dynamics finally concluding with the construction and testing of the target hypermedia system. Our ultimate objective is to use the RMD approach as the basis for the construction of computerized tools to support the design and development of hypermedia applications.

This chapter presents an edited transcript of an interview held in August 1993 with Brian Slater, a vice-president in the Global Bank at Chemical Bank. Slater is responsible for the firm's U.S.-based trading and global risk management systems. The interview was conducted in the context of Project 1990s -- the U.S. Council for International Banking's Study on Information Technology Investment and International Banking Performance. The purpose of the discussion was to develop background information on the issues the bank's senior management team faced in merging the trading and treasury functions of Chemical Bank and Manufacturers Hanover Trust. The new Global Bank faces challenges in the areas of global risk management, cost-effective delivery of in formation technology-based products, trading infrastructure application functionality gaps, and human resource management that can be best addressed by a shift to a new paradigm for trading and treasury application software development: reuse of object-oriented representations of the bank's systems. We now turn to the details of the discussion that develops these themes.

Maximizing business value of investments in hardware, software and telecommunications technologies that occur in the trading and treasury operations of an international bank requires senior management to evaluate the extent to which the technology infrastructure enables the bank to perform a number of key functions. These include: formulating effective trading strategies, pricing financial instruments accurately and rapidly, being able to respond to changing market conditions, processing transactions cost-effectively, resolving inquiries quickly, and moving to support emerging corporate treasury products. After a decade of rapid growth in investment levels, senior managers now emphasize refining, rationalizing and integrating trading and treasury technology architectures to support improved global financial risk management, better capital utilization, and higher transaction volumes. This chapter examines how senior managers can accomplish these goals by re-engineering pre-trade, trade execution and post-trade business processes. It presents a framework that utilizes basic concepts from management science and microeconomics to illustrate the variety of impacts that re-engineering can have on improving firm revenues and controlling or reducing costs. It also presents a series of managerial recommendations based on the framework.

Hypertext navigation alone is insufficient for efficient Information Retrieval (IR). Previous attempts to combine IR techniques with hypertext have been confined to the pre-authored structure of a document. In this paper we extend computer-science methods to synthesize a tailor-made hypertext document in response to each user's query. The synthesis technique can also be used to automatically create a pre-authored hypertext document according to an author's specifications.

During the last decade, the computational paradigms known as inflzcence diagrams and belief networks have become to dominate the diagnostic expert systems field. Using elaborate collections of nodes and arcs, these representations describe how propositions of interest interact with each other through a variety of causal and predictive links. The links are parameterized with inexact degrees of support, typically expressed as subjective conditional probabilities or likelihood ratios. To date, most of the research in this area has focused on developing efficient belief-revision calculi to support decision making under uncertainty. Taking a different perspective, this paper focuses on the inputs of these calculi, i.e. on the human-supplied degrees of support which provide the currency of the belief revision process. Traditional methods for eliciting subjective probability functions are of little use in rule-based settings, where propositions of interest represent causally related and mostly discrete random variables. We describe ratio-scale and graphical methods for (i) eliciting degrees of support from human experts in a credible manner, and (ii) transforming them into the conditional probabilities and likelihood-ratios required by standard belief revision algorithms. As a secondary contribution, the paper offers a new graphical justification to eigenvector techniques for smoothing subjective answers to pair-wise elicitation questions.

This paper addresses the problem of scheduling multiple time and priority sensitive tasks efficiently in an environment where the number of resources is limited and the resources have varying capabilities and restricted capacities. We use a help desk environment as our working model, however, the methodologv could also be adapted to a variety of job shop scheduling problems in general. We introduce a metric called priority time usage as a measure of task urgency and of schedule efficiency. We also introduce a method of considering user satisfaction in scheduling by utilizing fuzzy monotonic reasoning. We propose a methodology for implementing a heuristic genetic algorithm (GA) to accomplish the scheduling task. We discuss how such a system can use ongoing data about historical schedule performance to adapt and create progressively more accurate schedules in the future. We consider modifications to the scheduling approach which could allow for task inter-dependencies. We present an initiative user interface which we developed to aid help desk administrators in using the system. In addition to providing a front end to the SOGA system, the interface allows the user of the system to perform "what if” analysis with actual schedules. Lastly, we present preliminary assessments of the utility of both the optimization engine and the user interface.

What are the determinants of early interorganizational system (IOS) adoption? This paper focuses on a specific kind of IOS -- shared electronic banking networks -- and employs an economic approach that views adoption and diffusion in terms of cost and benefit. We attempt to identify firm characteristics that are likely to influence the perceived business value of network membership and develop specific hypotheses that can be tested empirically using historical data in a realistic setting. We undertake an exploratory econometric analysis of the adoption of Yankee 24, a large shared electronic banking network in the northeastern United States. Using Bass' analytical diffusion model, we categorize Yankee 24 network members into earlier and later adopters. Probit models are estimated to assess the impact of explanatory variables on shared electronic banking network adoption. The number of branch offices operated by a bank, its total demand deposits, and the proportion of its total deposits accounted for by demand deposits are found to be important predictors of earlier adoption. We find that the number of branch offices operated by a bank, a proxy for the size of its proprietary network, has a negative impact on early adoption, which contradicts the common wisdom that a large firm size is a prerequisite for adoption of technological innovations.

Methods for sound risk management are of increasing interest among Wall Street investment banking and brokerage firms in the aftermath of the October 1987 crash of the stock market. As the knowledge of advanced technology applications in risk management increases, financial firms are finding innovative ways to use them practically, in order to insulate themselves. The recent development in models, the software and hardware, and the market data to track risk are all considered advances in Risk Management Technology (RMT). -. These advances have affected all three stages of risk management: the identification, the measurement, and the formulation of strategies to control financial risk. This article discusses the advances made in five areas of RMT: communication software, object-oriented programming, parallel processing, neural nets and artificial intelligence. Systems based on any of these areas may be used to add value to the business of a firm. A business value linkage analysis shows how the utility of advanced systems can be measured to justify their costs.

Since the 1960s privacy advocates have relied on regulatory and legislative approaches to privacy protection in the United States, Canada and Europe. While important progress has been made in certain areas, there are large gaps and significant loopholes in existing legislation. I argue that a market-based approach to privacy protection would be far more effective and efficient in protecting individual information than current approaches.

A high-level simulation language based on temporal logic is described. The language combines a large set of temporal tenses and a rich class of high-level modeling primitives. Also an implementation of the language interpreter is presented. Finally. a real-world case study is described that shows how a programmer can develop structured, reliable, and well-maintainable simulation programs using the language.

In this paper we examined the differences and similarities in the human capital variables of male and female information system (IS) workers and the affect these differences had on job outcomes. The human capital variables studied included: age, education, organization and job tenure, and number of years in the IS occupation. We found that, even when controlling for the differences in human capital, women in IS still tended to be employed at lower levels of the organization, made less money, and had greater intentions to leave the organization.

Electronic markets use information technology to disseminate information on prices, quantities, and buyer and supplier identities. In spite of the recognized benefits of electronic markets, increased visibility and transparency may introduce imperfections, and create profitable opportunities to bypass markets that generates the information. In the U.S. securities markets, dissemination of market data has equipped several firms to develop competing, off-exchange trading mechanisms that rely on market price data, but whose transactions bypass the established market. Concern is rising that the growing volume of trading occurring away from the main market may reduce liquidity, and increase transactions costs. A simulation model of securities trading in a continuous auction market (similar to the market structure of the New York Stock Exchange) is used to examine the market quality effects of increasing levels of trading activity through an off-exchange dealer. Market characteristics, such as transactions costs, are measured as off-exchange trading increases from zero percent to 20 percent of the total trading volume. The results indicate that competition from an alternative trading venue reduces some trading costs borne by investors. Contrary to regulatory goals, however, off-market trading expands the role of profit-seeking dealers, and lowers the probability that some investors' orders will execute.

An open data network is one that allows firms to exchange information with outside entities, and a national data network is one that provides virtually universal access, interconnecting most businesses and private citizens in a country. Establishing a national data network is very costly and must be justified by its economic, service, and societal impact. France with its Teletel system is close to having a national data network, whereas the United States is not. Using data from 619 businesses in France and the United States we examine several claims: 1) open networks have improve firms' productivity, efficiency, and quality of service; 2) national open networks allow relationships among firms to be based on an electronic marketplace; and 3) national open networks differentially benefit small and medium sized firms. The results show that firms using open networks are more efficient and profitable and have more stable relationships with their customers. The natural advantage that large firms have to exploit new technology has been moderated in France, since small and medium-sized firms who used the national network have gained the same advantages as large firms. However, even in France, the use of open networks by the general population of firms is still relatively low, thus arguing for a long diffusion constant for the beneficial effects of open networks.

In this paper, we study the problem of discovering interesting patterns in large volumes of data. Patterns can be expressed not only in terms of the database schema but also in user-defined terms, such as relational views and classification hierarchies. The user-defined terminology is stored in a data dictionary that maps it into the language of the database schema. We define a pattern as a deductive rule expressed in user-defined terms that has a degree of certainty associated with it. We present methods of discovering interesting patterns based on abstracts which are summaries of the data expressed in the language of the user.

Numerous proposals for extending the relational data model to incorporate the temporal dimension of data have appeared in the past several years. These proposals have differed considerably in the way that the temporal dimension has been incorporated both into the structure of the extended relations of these temporal models, and consequently into the extended relational algebra or calculus that they define. Because of these differences it has been difficult to compare the proposed models and to make judgments as to which of them might in some sense be equivalent or even better. In this paper we define the notions of temporally grouped and temporally ungrouped historical data models and propose two notions of historical reIationa1 completeness, analogous to Codd's notion of relational completeness, one for each type of model. We show that the temporally ungrouped models are less expressive than the grouped models, but demonstrate a technique for extending the ungrouped models with a grouping mechanism to capture the additional semantic power of temporal grouping. For the ungrouped models we define three different languages, a temporal logic, a logic with explicit reference to time, and a temporal algebra, and show that under certain assumptions all three are equivalent in power. For the grouped models we define a many-sorted logic with variables over ordinary values, historical values, and times. Finally, we demonstrate the equivalence of this grouped calculus and the ungrouped calculus extended with a grouping mechanism. We believe the classification of historical data models into grouped and ungrouped provides a useful framework for the comparison of models in the literature, and furthermore the exposition of equivalent languages for each type provides reasonable standards for common, and minimal, notions of historical relational completeness.

Hypertext is one of those neat ideas in computing that periodically burst upon the scene, quickly demonstrating their usefulness and gaining widespread acceptance. As interesting, useful and exciting as hypertext is, the concept has certain problems and limitations, many of which are widely recognized. In this paper we describe what we call basic hypertext and we present a logic model for it. Basic hypertext should be thought of as a rigorously-presented approximation of first-generation hypertext concepts. Following our discussion of basic hypertext, we present our concept of generalized hypertext, which is aimed at overcoming certain of the limitations of basic hypertext and which we have implemented in a DSS shell called Max. We then present a logic model for browsing in generalized hypertext.

Re-engineering or business process redesign has become very popular. This paper presents a clear description of re-engineering and contrasts it with incremental change in systems. The paper also develops a framework for comparing two related systems. The framework is applied to a case study of the re-engineering of the Merrill Lynch Securities Processing System. This system features image processing, character recognition and extensive process redesign. The re-engineering effort has had a substantial impact on the firm.

The goal of this research is to provide hypermedia functionality to all information systems that interact with people. Hypermedia is a concept involving access to information, embodying the notions of context-sensitive navigation, annotation and tailored presentation. This paper presents the architecture of a system-level hypermedia engine, designed both to manage full hypermedia functionality for an information system and to bind interface-oriented front-end systems with separate computation-oriented back-end systems. The engine dynamically superimposes a hypermedia representation over a back-end application's knowledge components and processes. The hypermedia engine generates this representation using bridge laws, which capture the internal structure of client systems. Users access the application through its hypermedia representation. The paper also describes a set of minimal requirements for integrating the hypermedia engine with an information system. These guidelines apply to all integration efforts, not just that described here. Information systems will require some supplementary routines for the engine to manage hypermedia functionality for them. The more sophisticated and cooperative the information system, the higher the level of hypermedia support the engine will provide.

Software reuse in the presence of a repository and object-based CASE tool is likely to be "biased." Prior research (Banker, Kauffman and Zweig, 1991) showed that a developer will be: most likely to reuse her own objects; somewhat less likely to reuse objects developed by her project team members; and, even less likely to reuse objects stored in the repository, but developed elsewhere in the corporation. This paper characterizes this problem in terms of three familiarity biases: personal bias, project bias and time bias. In the presence of these biases it is appropriate to deploy tools that support the search for software reuse, so that they may be overcome. However, the tools that are chosen or created for this purpose must adequately treat the technical and cognitive fundamentals for individual developers, and recognize the organizational and economic perspectives of a firm that wishes to maximize the business value of its software development activities.

Many organizational problems can be decomposed into nearly independent subproblems the solution of which is the responsibility of independent agents. In this kind of work, which we call distributed work, the problems are only nearly independent since dependencies exist between the commitments required from each agent. As a consequence of these dependencies, the coordination problem becomes one of maintaining a consistent global solution in the face of the possibly conflicting activities of each agent. We define a normative model for coordination protocols that indicates the formal requirements for maintaining a globally consistent solution. The model identifies several properties that the protocol must enforce, namely serializability, atomicity, completeness, and soundness. We show that these properties are desirable in coordination protocols for distributed work problems.

Measurement of software development productivity is needed in order to control software costs, but it is discouragingly labor-intensive and expensive. Computer aided software engineering (CASE) technologies -- especially repository-based, integrated CASE -- have the potential to support the automation of this measurement. In this paper, we discuss the development of automated analyzers for function point and software reuse measurement for object-based CASE. Both analyzers take advantage of the existence of a representation of the application system that is stored within an object repository, and that contains the necessary information about the application system. We also discuss metrics for software reuse measurement, including reuse leverage, reuse value and reuse classification, that are motivated by managerial requirements and the efforts, within industry and the IEEE, to standardize measurement. The functionality and the analytical capabilities of state-of-the-art automated software metrics analyzers are illustrated in the context of an investment banking industry application, that is similar to systems deployed at the New York City-based investment bank where these tools were developed and tested.

Under circumstances where data quality may vary, knowledge about the potential performance of alternate predictive models can enable a decision maker to design an information system whose value is optimized in two ways. The decision maker can select a model which is least sensitive to predictive degradation in the range of observed data quality variation. And, once the "right" model has been selected, the decision maker can select the appropriate level of data quality in view of the costs of acquiring it. This paper examines a real-world example from the field of finance -- prepayments in mortgage-backed securities (MBS) portfolio management -- to illustrate a methodology that enables such evaluations to be made for two modeling alternative: regression analysis and neural network analysis. The methodology indicates that with "perfect data," the neural network approach outperforms regression in terms of predictive accuracy and utility in a prepayment risk management forecasting system (RMFS). Further, the performance of the neural network model is more robust under conditions of data quality degradation.

Methods for sound risk management are of increasing interest among Wall Street investment banking and brokerage firms in the aftermath of the October 1987 crash of the stock market. As the knowledge of advanced technology applications in risk management increases, financial firms are finding innovative ways to use them practically, in order to insulate themselves. The recent development in models, the software and hardware, and the market data to track risk are all considered advances in Risk Management Technology (RMT). These advances have affected all three stages of risk management: the identification, the measurement, and the formulation of strategies to control financial risk. This article discusses the advances made in five areas of RMT: communication software, object-oriented programming, parallel processing, neural nets and artificial intelligence. Systems based on any of these areas may be used to add value to the business of a firm. A business value linkage analysis shows how the utility of advanced systems can be measured to justify their costs.

The goal of this research is to provide hypermedia functionality to all information systems. In this paper I present the architecture of a system-level hypermedia engine, designed both to manage fill hypermedia functionality for an information system and to bind interface-oriented “front-end" systems with separate computation-oriented "back-end "systems. The engine dynamically superimposes a hypermedia representation over a back-end application's knowledge components and processes. I then describe a set of minimal requirements for integrating the hypermedia engine. The more sophisticated and cooperative the information system, the higher the level of hypermedia support the engine will provide.

The Dempster Shafer theory of evidence concerns the elicitation and manipulation of degrees of belief rendered by multiple sources of evidence to a common set of propositions. Information indexing and retrieval applications use a variety of quantitative means - both probabilistic and quasi-probabilistic - to represent and manipulate relevance numbers and index vectors. Recently, several proposals were made to use the Dempster Shafer model as a relevance calculus in such applications. The paper provides a critical review of these proposals, pointing at several theoretical caveats and suggesting ways to resolve them. The methodology is based on expounding a canonical indexing model whose relevance measures and combination mechanisms are shown to be isomorphic to Shafer 's belief functions and to Dempster's rule, respectively. Hence, the paper has two objectives: (i) to describe and resolve some caveats in the way the Dempster Shafer theory is applied to information indexing and retrieval, and (ii) to provide an intuitive interpretation of the Dempster Shafer theory, as it unfolds in the simple context of a canonical indexing model.

This paper argues that information technology can have a significant impact on organizational flexibility. Information technology (IT) contributes to flexibility by 1) changing the nature of organization boundaries and the time at which work takes place and 2) altering the nature and pace of work. IT also has important second and third order impacts on organizations and industries. The paper presents examples to illustrate the impact of information technology on two industries and three companies. The paper concludes that management should consider the use of information technology to increase flexibility and should predict the higher order impact of their actions on the organization and their industry.

Traditional management accounting systems are limited in their ability to provide profitability information relevant to management decisions. The problems of inadequate profitability measurement are intensified in today's business environments, where changing margins due to deregulation and new entrants, new products with unknown costs, and customer sophistication in locating low-cost providers often combine to leave unprepared firms with growing numbers of loss-making client relationships. In response, firms in a number of service and manufacturing industries are experimenting with new methods for measuring performance, and are implementing these techniques using information systems. The collection and analysis of information on the profitability of customer relationships enables managers to identify and defend their most attractive market segments, and to turn loss-making accounts into profitable ones. The London-based securities house, BZW, developed BEATRICE, an innovative information system that combines activity-based accounting principles and a model of customer profitability to make an income assignment to each of the 6,000 trades the firm makes in a day. The system's value is considerable, and can be evaluated by using industry performance benchmarks, and by comparing management decision-making with the currently available information to what was possible with previous data.

The London Stock Exchange (LSE) faces rising pressure in its efforts to maintain its position as a favored market of institutional fund managers and professional investors. Customers are satisfied with the current state of the LSE market, but member firms are pricing institutional brokerage and market making services below economic cost. The LSE's position will be significantly damaged when the effective subsidy ends, and commissions and dealing spreads reflect the costs incurred by members firms. Competition in the supply of trading services has increased, and a range of alternative, off-exchange trading systems could draw order flow away from the Exchange market. These trading mechanisms provide order matching, crossing of basket and portfolio trades, and reduce investors' commissions and trading spread costs. Fund managers in the U.S. are using the systems more actively, and the result has been an erosion of the position of the New York Stock Exchange. The LSE's customers are also using an expanding range of portfolio management techniques, many of which require low-cost trading, and do not demand immediate order execution, as traditionally provided by London's market makers. The Exchange needs to respond to the changes in fund managers' demand for trading services, and to the growing competition in the supply of off-exchange trading services. Enhancements to the existing LSE market structure are the best response to these threats.

Models of bibliographical Indexing concern the construction of effective keyword taxonomies and the representation of relevance between document s and keywords. The theory of evidence concerns the elicitation and manipulation of degrees of belief rendered by multiple sources of evidence to a common set of propositions. The paper presents a formal framework in which adaptive taxonomies and probabilistic indexing are induced dynamically by the relevance opinions of the library's patrons. Different measures of relevance and mechanisms for combining them are presented and shown to be isomorphic to the belief functions and combination rules of the theory of evidence. The paper thus has two objectives: (i) to treat formally slippery concepts like probabilistic indexing and average relevance, and (ii) to provide an intuitive justification to the Dempster Shafer theory of evidence, using bibliographical indexing as a canonical example.

We discuss different techniques for representing infinite temporal data. There are two basic approaches: A procedural description, as used in production systems, and represented, in this paper, by a version of Datalog. The second approach is a more declarative method, using some form of temporal logic programming. We examine several versions of each approach, and compare their expressive power, i.e., what temporal data each formalism can capture.

This paper argues that information technology can have a significant impact on organizational flexibility. Information technology (IT) contributes to flexibility by 1) changing the nature of organization boundaries and the time at which work takes place and 2) altering the nature and pace of work. IT also has important second and third order impacts on organizations and industries. The paper presents examples to illustrate the impact of information technology on two industries and three companies. The paper concludes that management should consider the use of information technology to increase flexibility and should predict the higher order impact of their actions on the organization and their industry.

This paper reports on a multi-year field study conducted at the First Boston Corporation to investigate the performance of its object and repository-based integrated computer aided software engineering (I-CASE) tool called High Productivity Systems (HPS). We present productivity and software reuse results for: * three closely monitored small-scale I-CASE pilot projects; and, * twenty large-scale investment banking projects built during the first two years of the I-CASE tool's deployment. Our results offer perspective on the potential for order of magnitude gains in software development productivity when reusable software development approaches are applied. They also suggest the need for new measurement and management approaches to foster improved software development performance in the I-CASE world.

Customer access information technology offers great potential benefits to both the organization and its customers. However, managers lack methods for evaluating these technologies and determining whether conditions exist for success. We propose a model from Marketing Science which captures factors influencing the success of a CAIT. The factors include value platform features - features of the CAIT, and location strategy features - features of the environment into which the CAIT is deployed. This model is applied to home banking, providing insights as to types of systems most likely to be successful.

Achieving alignment between the goals of the information systems (IS) function and the organization as a whole remains a top priority. A perceptual instrument is described that sets out to measure this alignment. It allows organizations to monitor the alignment and effectiveness of their IS function over time and to compare their situation with others. Largescale surveys of different industry sectors and more extensive studies of individual companies in the United States and South Africa have been undertaken using the instrument. The results are used to evaluate the reliability and validity of the instrument. Several hypotheses regarding alignment are tested. The results suggest that the degree of alignment between the importance and performance of specific aspects of IS influences overall perceptions of IS success. This applies to assessments of both IS staff and users. It is also found that IS staff and users are mostly in agreement about the importance of different aspects of IS and the success with which they are being performed, but the extent of this agreement is not a predictor of overall success. Conclusions are drawn regarding the link between alignment and effectiveness of the IS function and recommendations are made for researchers and practitioners.

Information Systems Working Papers Series

As organizational environments become more turbulent and complex and as uncertainty about the future increases, reliance on quantitative decision-making approaches for strategic planning becomes less appropriate. Scenario analysis can be an effective qualitative technique for enhancing strategic planning. Typically, scenarios are presented as alternate futures. Some theorists, however, have suggested that forward and backward thinking are different cognitive processes. In this study, we investigate the effect of presenting scenarios retrospectively; that is, as if future events had already happened. A repeated measures laboratory study compared the performance of professional planners (n = 64) using prospective and retrospective scenarios in two simulated business planning tasks. Measures consisted of objective factors (number of individual planning statements and number of monitoring statements), subjective factors (quality based on an 18-question rating instrument), and subject attitudes concerning their experience with the two treatments. Results suggest that use of retrospective scenarios do increase the number of planning statements. In addition, plans prepared using retrospective scenarios were rated higher overall than those prepared with prospective scenarios. Moreover, evidence emerged that a subset of subjects were better able to make use of the retrospective technique, suggesting that selection along with training may improve planning performance.

The computer industry is characterized by a number of successful and unsuccessful start-ups. Past studies of innovation have presented "market pull" and "technology push" models to describe the development of new products. This paper proposes a synthesized model of concurrent technology and market strategy development. This model is illustrated with a case study of a successful "medium-technology” start-up company. For this start-up firm, market needs drove the development of technology while the technology enabled the firm to achieve its market strategy.

A method of adding time to active databases is described in this paper. This is achieved by incorporating operators of temporal logic, temporal actions, and different temporal clauses into the Event-Condition-Action model of a rule. In addition, a temporal recognize-act cycle is described and new temporal conflict resolution strategies are proposed. A conflict avoidance strategy for temporal rules is also described.

Numerous proposals for extending the relational data model to incorporate the temporal dimension of data have appeared in the past several years. These proposals have differed considerably in the way that the temporal dimension has been incorporated both into the structure of the extended relations of these temporal models, and consequently into the extended relational algebra or calculus that they define. Because of these differences it has been difficult to compare the proposed models and to make judgments as to which of them might in some sense be equivalent or even better. In this paper we define the notions of temporally grouped and temporally ungrouped historical data models and propose two notions of historical relational completeness, analogous to Codd's notion of relational completeness, one for each type of model. We show that the temporally ungrouped models are less powerful than the grouped models, but demonstrate a technique for extending the ungrouped models with a grouping mechanism to capture the additional semantic power of temporal grouping. For the ungrouped models we define three different languages, a temporal logic, a logic with explicit reference to time, and a temporal algebra, and show that under certain assumptions all three are equivalent in power. For the grouped models we define a many-sorted logic with variables over ordinary values, historical values, and times. Finally, we demonstrate the equivalence of this grouped calculus and the ungrouped calculus extended with the proposed grouping mechanism. We believe the classification of historical data models into grouped and ungrouped provides a useful framework for the comparison of models in the literature, and furthermore the exposition of equivalent languages for each type provides reasonable standards for common, and minimal, notions of historical relational completeness.

This paper presents a logic modeling exercise in which we develop test and implement a logic model for a test editor and use it to test existing test editing software. We begin by presenting a first order Horn logic axiomatization of a text editor by providing domain equations for the primitive operations insert, delete and character retrieval. We show that this logic model captures the essential aspects of the text editing task and how more complex features are built using these primitives. We discuss possible implementations and conclude that any operational semantics - the set of algorithms that perform the task - must be strongly related to the logic model we present. In other words, each operational semantics constitutes a model of the logic theory Next, we illustrate the usefulness of the model by implementing a basic text editing system and testing the correctness of an existing text editor. We conclude by describing how we are integrating these modeling techniques into a larger and more complex knowledge-based system.

Data quality has been shown to be a major determinant of the value of systems that utilize input data feeds and transform them into valuable information under a variety of business contexts. For this study, we have chosen a financial risk management context to investigate the relationship between data quality and value of risk management forecasting systems. Three attributes of data quality, frequency, response time, and accuracy, along with the cost of data are considered. Joint impacts of attributes are also considered. It is shown that an increase in report frequency results in an increase in the utility of a risk management forecasting system, but this increase is limited by the responsiveness of the hedging scheme. Frequency is shown to improve the utility of the forecasting systems in two ways: First, an increase in frequency pushes the predicted states closer to the actual states and second, an increase in frequency causes the reliability of the forecasting model to increase. A delay in response time of reports is predicted to have a greater impact on utility for high frequency reports than for low frequency reports. Finally, data inaccuracies are recommended to be the first concern of a portfolio manager before an attempt is made to increase the reporting frequency.

Neural networks offer an approach to computing which - unlike conventional programming - does not necessitate a complete algorithmic specification. Furthermore, neural networks provide inductive means for gathering, storing, and using, experiential knowledge. Incidentally, these have also been some of the fundamental motivations for the development of decision support systems in general. Thus, the interest in neural networks for decision support is immediate and obvious. In this paper, we analyze the potential contribution of neural networks for decision support, on one hand, and point out at some inherent constraints that might inhibit their use, on the other. For the sake of completeness and organization, the analysis is carried out in the context of a general-purpose DSS framework that examines all the key factors that come into play in the design of any decision support system.

Traditionally, software management has focused primarily upon cost control. Today, with the emerging capabilities of computer aided software engineering (CASE) and corresponding changes in the development process, the opportunity exists to view software development as an activity that creates reusable software assets, rather than just expenses, for the corporation. With this opportunity comes the need to monitor software at the corporate level, as well as at that of the individual software development project. Integrated CASE environments can support such monitoring. In this paper we propose the use of a new approach called repository evaluation, and illustrate it in an analysis of the evolving repository-based software assets of two large firms that have implemented integrated CASE development tools. The analysis shows that these tools have supported high levels of software reuse, but it also suggests that there remains considerable unexploited reuse potential. Our findings indicate that organizational changes will be required before the full potential of the new technology can be realized.

One fundamental requirement in the expected utility model is that the preferences of rational persons should be independent of problem description. Yet an extensive body of research in descriptive decision theory indicates precisely the opposite: when the same problem is cast in two different, but normatively equivalent, "frames," people tend to change their preferences in a systematic and predictable way. In particular, alternative frames of the same decision tree are likely to invoke different sets of heuristics, biases, and risk-attitudes, in the user's mind. The paper presents a computational model in which decision-trees are cast as attributed graphs, and reframing operations on trees are implemented as graph-grammar productions. In addition to the basic functions of creating and analyzing decision-trees, the model offers a natural way to define a host of "debiasing mechanisms" using graphical programming techniques, Some of these mechanisms have appeared in the decision theory literature, whereas others were directly inspired by the novel use of graph grammars in modeling decision problems.

Techniques for analysis and validation of software requirements specifications written in the knowledge representation language Templar are presented. Templar specifications are analyzed in terms of ambiguity, non-minimality, contradiction, incompleteness, and redundancy. Since Templar is a powerful knowledge representation language supporting a rich set of modeling primitives, it is difficult to reason directly on Templar specifications. To solve this problem, Templar specifications are mapped into equivalent temporal logic programs which are analyzed in terms the criteria listed above. However, it is hard to reason about Templar specifications because some of the criteria cannot be formally proven, and the verification of other criteria constitute undecidable or intractable problems. To overcome these difficulties, we consider a set of tractable conditions for each criteria, which serve as "alarms" for the user. If a condition is violated then it means that the specification either definitely has or potentially can have a problem. Furthermore, the user is notified about the source and the nature of the problem in certain cases.

Several researchers characterized the activation functions under which multilayer feedforward networks can act as universal approximators. We show that all the characterizations that were reported thus far in the literature ark special cases of the following general result: a standard multilayer feedforward network can approximate any continuous function to any degree of accuracy if and only if the network's activation functions are not polynomial. We also emphasize the important role of the threshold, asserting that without it the last theorem doesn't hold.

Existing output measurement metrics for cost estimation and development productivity need to be re-examined to determine their performance in computer aided software engineering (CASE) development environments. This paper critiques and empirically evaluates four approaches to the measurement of outputs. Two of the metrics, raw function counts and function points, are based on the function point analysis methodology pioneered by Albrecht and Gaffney at IBM (ALBR83). The second two, object counts and object points, are based on a new approach -- object points analysis -- that is introduced here for the first time. The latter metrics are specialized for output measurement in object-based CASE environments that include a centralized object repository. Estimation results for nineteen large-scale CASE projects show that the new metrics have the potential to yield as accurate, yet easier to obtain estimates than function points-based measures.

One of the primary methods of instruction in business education is the case. However, some of the basic learning goals associated with case instruction are compromised by its form of presentation. We have used new forms of media, involving computing and communication, to build a novel learning environment, the Living Case, which flexibly and interactively presents cases along with dynamic, on-going feedback to students while they work. Key to providing meaningful assistance daring case analysis is the ability to model and interpret student behavior. Several investigations were conducted in order to understand the process of case analysis. Case analysis is characterized as a problem solving activity driven by comprehension and reasoning operators. Twelve hours of protocols are analyzed using "retelling profiles" as an interpretation mechanism for further specifying the operators involved in analyzing a case. Retelling profiles are visual time plots of the activities undertaken in a reading task. Our preliminary results suggest a deeper structure to case analysis which is common across business disciplines, cases, and analysts, and therefore implementable in a system like Living Case. Differences between the analysis strategies of experts and novices are formalized in terms of the experts' use of "templates" of typical company behaviors. This provides the basis for building mechanisms to instruct and re-orient case analysts using the Living Case system. The Living Case system is described along with some of the insights gained during its construction. Future research directions and instructional uses of the system are also discussed.

In this paper we discuss the application of hypertext valuation links to decision support for business problems. Valuation links enable us to relate hypertext link traversal to computation in a way that affects the contents of a hypertext node while retaining the "browsing metaphor'' of hypertext. This helps to support quantitative or qualitative reasoning about business problems when described in terms of hypertext nodes that are computational in nature. We illustrate these ideas in the domain of securities analysis and portfolio management, where a "buy side" portfolio manager may need to clarify his understanding of the basis of a "sell side" securities analyst's recommendations about securities that are candidates for inclusion in a portfolio.

In design, inferring structure from function is a combinatorial generate-and-test problem. Existing methods use prestored domain-specific partial configurations to constrain the generator. We have found that for certain types of economic and physical systems consisting of two-terminal components connected in parallel, it is fruitful to specify function in terms of desired behavior, and to identify sets of components whose resultant behavior matches that desired behavior. In this paper, we present two synthesis operators called stretch and steepen that operate on qualitatively specified piecewise linear functions that characterize the behavior of components. We are currently applying this model to the domain of financial hedging, where behaviors of the components (stocks, bonds, options, etc.) are specified in terms of two-dimensional piecewise linear relationships, and the goal is to synthesize these to produce a constrained behavior in response to uncontrollable events.

In this paper we study the integration of Model Management and Hypertext systems to produce a Hyper Model Management System (HMMS). Model Management Systems constitute a class of software that is designed to support the construction, storage, retrieval, and use of models in the context of decision support systems (DSS). Hypertext systems allow users to split information into data fragments which the user can browse to find information by taking non-linear paths in computer based texts. It has been suggested that DSSs should be conceived as environments which support decision making. We support the view that such environments can be readily provided for the subtask of model management by hypertext systems. The different kinds of model knowledge can be captured within different types of nodes and the relationships among these can be maintained by hypertext links. In this paper we describe some aspects of model management where hypertext will have a significant impact. However, plain hypertext is ineffective in dealing with the dynamic nature of information in model management tasks where data is revised, models executed, and reports are created on the fly. Dynamic domains require dynamic hypertexts. In this paper we also study the requirements for dynamic hypertexts. These can be satisfied within the class of generalized hypertext systems by using special hypertext nodes and links which we describe. We explore different architectures to integrate MMS and Hypertext systems to obtain HMMSs. This paper emphasizes the need for a shift to integrated Model Management environments and proposes hypertext as an integrating technology.

The paper illustrates ‘business value linkage impact analysis’, a new method for measuring the business value of information technology (IT), in the context of a case study of electronic banking operations at Meridian Bancorp, a large commercial bank. Management science models were constructed to gauge the impact of automated teller machines (ATMs) on branch teller labour productivity and retail deposit market share, and the potential for substitution of labour by ATMs is shown. Econometric estimation of the models yielded the following results: the efficiency of teller labour was found to decline in the presence of a branch ATM; a bank's ATM network decision was shown to be an important determinant of the relative size of the retail deposit market it could capture in south-east Pennsylvania; membership in the regionally dominant MAC ATM network leveraged retail deposit market share when a clear majority of local branches and ATMs were members of a regionally smaller, competing network: and high-density ATM deployment did not lead to increases in the overall size of the deposit market.

In this paper we discuss the application of hypertext valuation links to decision support for business problems. Valuation links enable us to relate hypertext link traversal to computation in a way that affects the contents of a hypertext node while retaining the "browsing metaphor” of hypertext. This helps to support quantitative or qualitative reasoning about business problems when described in terms of hypertext nodes that are computational in nature. We illustrate these ideas in the domain of securities analysis and portfolio management, where a "buy side" portfo1io manager may need to clarify his understanding of the basis of a "sell side” securities analyst's recommendations about securities that are candidates for inclusion in a portfolio.

This paper presents a new approach to measuring the input productivity gains from information technology (IT) in complex managerial environments. The approach is illustrated in the context of a study of a pilot deployment at Hardee's Inc. of a new cash register point-of-safe and order-coordination technology called “Positran." The method employs data envelopment analysis (DEA) and nonparametric production frontier hypothesis testing to determine whether the performance of restaurants that have deployed Positran is better, on average, than for those that have not. The design of the study is of special interest because it approximates a controlled experiment. Our results show that Positran helped to reduce input materials costs, since restaurants that deployed the technology were less likely to be inefficient It is further possible to characterize the class of restaurants for which the relationship holds. Operational efficiency measures such as the ones we have developed provide managers with the opportunity to implement deployment strategies for new ITs in order to maximize value.

One of the primary methods of instruction in business disciplines is the case. However, in its current form of written presentation, some of the basic learning goals associated with case instruction are compromised. We have used new forms of media involving computing and communication to build a novel learning environment, the Living Case, which flexibly and interactively presents cases along with provide dynamic, on-going feedback to students analyzing a case. In our research we have formalized the process of analyzing a case in order to recognize and interpret student analysis behavior so that provide relevant assistance can be provided. Case analysis is characterized as a problem solving activity driven by comprehension and reasoning operators. A search for these operators led us to build an inventory of reading activities. Twelve hours of protocols are analyzed using "retelling profiles" as an interpretation mechanism. Retelling profiles are visual time plots of the activities undertaken in a reading task. Our preliminary results suggest a deeper structure to case analysis which is common across business disciplines, cases, and individuals. Differences between the analysis strategies of experts and novices are also formalized. Finally, an expert's analogical reasoning strategy using task-specific knowledge encoded as "templates" is identified as a major contributor to their efficiency in solving cases. Templates of typical company situations and responses are triggered early in the analysis process, and subsequent data gathering and reasoning is directed by an attempt to apply the template to the case situation.

Behavior of relational databases is studied within the framework of Relational Discrete Event Systems (RDESes) and Models (RDEMs). Three behavior specification methods based on production systems, recurrence equations, and Petri nets are defined and their expressive powers are compared. Production system RDEM is extended to support non-determinism, and various deterministic and non-deterministic production system interpreters are introduced and formally compared in terms of their expressive power. It is shown that the parallel deterministic interpreter has more expressive power than other interpreters including an OPS5-like interpreter. Since it is also parallel, this makes the parallel deterministic interpreter a very attractive interpreter for production systems.

When dealing with nearly decomposable problems such as those described by Simon (1973), the problem components may be worked on by different problem solvers that are spatially and temporally separated, with each problem solver constrained by assumptions it makes about the activities and choices of other problem solvers, that is by partial knowledge of the global problem. There are advantages to maintaining multiple solutions locally for as long as possible, even though a single final solution is desired. When it becomes less desirable to retract certain assumptions, these become constraints for other problem solvers and can be communicated to them via a truth maintenance system. We describe an extended architecture for an ATMS for these kinds of decomposable problems

An examination of code reuse at a large financial institution yields insights into the process of code reuse. The software development environment -- based on an integrated CASE system -- was designed to support code reuse, but at the end of its first two years we find that programmers are not taking full advantage of the reuse opportunities which the CASE environment provides, The organization has provided technical support for code reuse, but has not made organizational adjustments, and the technical solution alone does not suffice. We also review an existing economic model of CASE development performance that incorporates code reuse, suggesting refinements that are based upon our observations. Finally, we draw some conclusions about steps that managers can take to promote code reuse.

We advocate the use of temporal logic instead of the first-order logic in rules of knowledge-based simulation systems. We argue that this provides several advantages that will be discussed in the paper. We show how temporal logic is used in simulation by considering language PTL based on temporal logic programming.

This paper reports on an effort to integrate symbolic and mathematical models to tailor the optimal output of an operations research model to the particular domain of a decision maker. AESOP combines the Black-Scholes model of stock options pricing with an expert system; the integrated model is designed for use by an options specialist on the American Stock Exchange. The specialist makes a number of adjustments to the output of the mathematical model; the purpose of the symbolic model is to make as many of these modifications as possible automatically. The paper reports on the development and structure of AESOP and presents data on its use.

We define a temporal algebra that is applicable to any temporal relational data model supporting discrete linear bounded time. This algebra has the five basic relational algebra operators extended to the temporal domain and an operator of linear recursion. We show that this algebra has the expressive power of a safe temporal calculus based on the predicate temporal logic with the until and since temporal operators. In [CrC189], a historical calculus was proposed as a basis for historical relational completeness. We propose the temporal algebra defined in this paper and the equivalent temporal calculus as an alternative basis for temporal relational completeness.

This paper presents a new approach for evaluating the quality of managerial choices in information technology (IT) deployment. The approach involves measuring the extent to which deployment sites perform in accordance with the firm's objectives, given the constraints of their competitive environment. Our method is to model environmental descriptors as inputs to a production process that yields business outputs. This production process is then evaluated via standard productivity assessment methods to obtain "competitive efficiency” scores. Interpreting why different deployment sites exhibit different levels of competitive efficiency involves estimating regression models in which competitive efficiency scores are the dependent variables and management's IT design choices are the independent variables. Such measurement and interpretative methods provide managers with new tools to improve their IT location and design decisions. Our framework is illustrated in the context of automatic teller machine (ATM) deployment.

This paper makes the business case for automating the collection of software metrics for gauging development performance in integrated computer aided software engineering (CASE) environments that are characterized by an object-oriented development methodology and a centralized repository. The automation of function point analysis is discussed in the context of such an integrated CASE environment (ICE). We also discuss new metrics that describe three different dimensions of code reuse -- leverage, value and classification -- and examine the p,ossibility of utilizing objects as means to estimate software development labor and measure productivity. We argue that the automated collection of these software metrics opens up new avenues for refining the management of software development projects and controlling stra-egic costs.

This study investigates the relationship between system success as operationalized by user information satisfaction (UIS) and various economic measures of firm performance. The findings indicate a significant positive but complex relationship between firm performance and UIS. In particular, we found that it is inappropriate to aggregate UIS scores across individuals within a firm. The CEO, Controller, and Production Manager within a firm tended to have quite different UIS scores, resulting in low interrater reliabilities. We also found, that the association between a respondent's UIS score and the measures of firm performance depended heavily on the position of the respondent and the particular performance measure employed.

In this paper we present a general representation for constraint satisfaction problems (CSP) and a - framework for reasoning about their solution that unlike most constraint-based relaxation algorithms. stresses the need for a "natural" encoding of constraint knowledge and can facilitate making inferences for propagation, backtracking, and explanation. The representation consists of two components: a generate-and-test problem solver which contains information about the problem variables, and a constraint-driven reasoner that manages a set of constraints, specified as arbitrarily complex Boolean expressions and represented in the form of a constraint network. This constraint network: incorporates control information (reflected in the syntax of the constraints) that is used for constraint propagation: contains dependency information that can be used for explanation and for dependency-directed backtracking; and is incremental in the sense that if the problem specification is modified, a new solution can be derived by modifying the existing solution.

Information Systems Working Papers Series

MCADSS is a multi-criteria allocation decision support system for assisting in the task of partitioning a set of individuals into groups. Based upon multiple criteria, MCADSS’s goal is to maximize the diversity of members within groups, while minimizing the average differences between groups. (The project may be viewed from several perspectives: as a multi-criteria decision making problem, as a "reverse" clustering problem, or as a personnel assignment problem). The system is currently being used to allocate MBA students into sections and study teams at INSEAD, a leading European business school. This paper describes the rationale for MCADSS, design criteria, system methodology, and application results. It also suggests how the approach outlined here might be used for further applications.

MESS is an expert system to support a decision at the American Stock Exchange to open an investigation of insider trading. The system was developed with three experts and extensive prototyping. MESS illustrates multiple roles for an expert system; it contributes to decision making by seeing that all relevant factors are considered, stimulating discussion of important criteria, and providing an audit trail for regulatory review The system also serves a training role for a position with historically high turnover.

Experienced data entry operators participated in an experiment to evaluate the effects of the error notification timing on error correction performance irt the data entry task. Three timing strategies were studied: immediate error notification, notification at the end of a field, and notification at the end of a physical line. Based on a model of data entry and a model of the correction process hypotheses were developed to predict operator performance. It was conjectured that operators treat individual fields as separate units tasks and that interrupting within a task would be more disruptive than interrupting between tasks. The results of the study indicated that the error rates during correction were smaller for the end of line treatment although the time to complete the correction was longer for this treatment than the other two. Performance was essentially the same for those operators interrupted who were interrupted immediately and those who were interrupted at the end of a field.

Information Systems Working Papers Series

The conceptual design of user interfaces focuses on arriving at a specification of the structure of the dialog, independent of any particular implementation approach. There is common agreement as to the importance of this activity to both IS professionals and end-users, but few -- if any -- modeling methods were developed to specifically support the process of conceptual design, and the usefulness of such methods has not been adequately addressed. This paper introduces the Dialog Charts (DCs), and documents a preliminary examination of their perceived usefulness by designers of user/system interaction who actually used them. The DCs yield high level dialog schemas that are abstract enough to support the conceptual design of dialog control structures. In a uniform diagramming framework they combine the concept of dialog independence, distinguish between the dialog parties, provide for hierarchical decomposition and enforce a structured control flow. The usefulness of the DCs has been studied empirically in a qualitative inquiry. Recalled experiences of designers were captured and analyzed to ascertain the concept of usability, as well as assess the usability of the DCs. Usability has emerged from this study as a set of 38 concerns that operationalizes the broader aspects of purpose of use, design stage, impact on product structure, impact on design process, and attitudinal patterns. In general, the Dialog Charts were found by these dialog designers to be a useful, exhibiting the essential attributes of tools for conceptual modeling.

We develop a framework based on "project networks" and net present value analysis in order to help managers evaluate investments in infrastructural and strategic information systems that require significant amounts of time and money to implement. The framework, which we term "value networks", supports decomposition of investment projects into separate increments, offers a means to represent crucial dependencies that affect the creation of IT business value, and provides a basis for developing a measurement methodology which can be used from the planning through the implementation stages. This enables the user to identify the implications of choosing among different implementation schedules. We illustrate these ideas by examining how our framework can used to evaluate investments in Electronic Data Interchange (EDI) and Cash Management Systems (CMS).

Knowledge Acquisition is widely recognized as the single major bottleneck in the commercialization of Expert Systems technology. The typically ad-hoc choice of techniques for eliciting and representing expert knowledge, makes Expert Systems development expensive and prone to failure. Arguments have been made in the Knowledge Acquisition literature for performing an epistemological or "knowledge-level" analysis to "structure" the knowledge elicitation process. The need of the hour is for an empirical evaluation of these claims. In this paper, we present the results of a study that evaluates an approach to Structured Knowledge Acquisition, that is based on analyzing expert behavior using generic problem-solving tasks. Data from a large Expert Systems project currently nearing completion, has been used for the study.

Current planning and control procedures for large-scale software projects are not sufficiently equipped to deal with changing or imprecise requirements, resource breakdowns, unexpected delays, etc. We propose a solution for managing change in projects, based on a semantic model of the software design and development processes. At the heart of this technique is the formation of islands of project knowledge in a way that facilitates dealing with most design and plan revisions locally. A protocol for interactive change management is presented that advocates need-based formation of coalitions between islands as a means for graceful degradation in the place of strict hierarchical control. The results of initial empirical investigations of the usability of the approach and plans for its continuing evaluation are also reported.

Recent research on information technology (IT) value has focused on examining new theoretical bases from which to construct robust valuation methods and models. This paper considers two literatures which previously have not been explored in this context: research on behavioral decision making and information presentation effects. We begin by identifying a typology of potential cognitive biases and heuristics which may enhance or suppress IT value when workstations are used to provide decision support. To illustrate, we examine how these effects may become operative in screen-based securities and foreign exchange trading activities, where designers can choose among information presentation formats which support trader decision making. We adapt a recent model by Kroeck, Kirs and Fiedler (1989) to identify where and how information effects, heuristics and biases come into play in the trading environment. Our investigation concludes that managerial recognition of the potential value tradeoffs associated with alternative trading workstation window designs is an important concern for fine-tuning trading decision support systems. In this way, the "business value linkage" between trading workstation investments and the returns they provide can be more fully understood.

Empirical survey research has been extremely popular in the information systems field. However, survey researchers encounter a number of methodological problems in conducting their studies. This paper presents a model of the research process and uses it to organize a discussion of the difficulties encountered in doing IS survey research. The paper concludes with suggestions on how to improve the quality of IS survey research and improve the information systems field at the same time.

We present expert system (ES) and Integer Programming (IP) formulations of an NP-complete constraint satisfaction problem (CSP). The problem involves generating a plan for assigning faculty to courses given a variety of constraints and preferences and other tentative data. The expert system consists of a heuristic rule-based problem solver and a truth maintenance system. The IP model consists of about 700 zero-one decision variables and 300 constraints. We describe and contrast the expert system and IP models in terms of behavior, quality of results, and computational performance. We find that the expressiveness of the IP model is hampered by its single objective function, inability to encode various types of complex preferences, the lack of useful output when it fails to find a feasible solution, and a general lack of control over inference. It is also difficult to make incremental revisions to the plan produced by the IP model. In contrast, the truth maintenance system maintains justifications for assignments, which makes it possible to reason about incremental modifications to a plan. In terms of performance, we found that whenever the IP approach finds a solution, it does so quickly using the Pivot and Complement heuristic of Balas and Martin (1980). The branch and bound always failed to find a feasible integer solution when the heuristic failed to find one.

Much of transaction processing involves classification, that is, the categorization of inputs into outputs based on various tests. In Artificial Intelligence (Al), classification systems are generally represented in terms of AND/OR graphs. Such graphs are collections of production rules that capture declaratively the logic of an application domain. If one views a transaction processing system as a classification system, it becomes natural to represent it in terms of an AND/OR graph. In this paper, we present an interpretation of dataflow diagrams used in Structured Analysis as AND/OR graphs. By examining the dataflow diagrams, production rules capturing application-specific knowledge can be constructed. This interpretation has two implications: 1) production rules can be used to unify analysis and design since the same data structure (the rule) is used for both purposes, and 2) the resulting design can be simulated for purposes of explanation and what-if analysis. We also discuss some of the general pros and cons of production systems as they pertain to systems analysis and design.

The purpose of this paper is to develop a general model of the process by which large organizations develop information technology over long periods of time. A special focus of the paper concerns the question how "social impacts of computers" are produced by management decisions, organizational exigencies, accidents, and environmental forces. The model is clearly situated in the broader behavioral literature on organizations and organizational innovation. Major streams of behavioral research and assumptions are reviewed. The model assumes an eclectic position: organizational innovation results from both internal institutional factors as well as powerful environmental forces. "Social impacts of computers" filter out from a reasonably complex interaction between the organization and the environment. Our goal from the outset was to develop a general model of information technology development which was not a "special" purpose, narrowly framed theory typical of prior management information system research. In addition, we hope to set straight popular misconceptions created by vendors, consultants, and others concerning the question, how do computers "impact" organizations.

Developing robust and refined measures to quantify strategic impacts is a major challenge facing researchers who seek to improve methods for information technology (IT) investment evaluation. This paper presents a means to test for parameter heterogeneity in a model which quantifies the strategic contribution of IT. An iterative "jackknife" procedure is used to diagnose if different local competitive and demographic conditions present in branch banking enhance or suppress leverage on deposit market share associated with membership in an automated teller machine (ATM) network. The results are validated using correlation analysis and re-estimating partitioned data sets for a market share model developed by Banker and Kauffman (1988). Overall, the results suggest this new approach will be useful for managers who need to evaluate similar ITS which operate in different environments.

Information Systems Working Papers Series

Information Systems Working Papers Series

Constraints, in various forms, are ubiquitous to design problems. In this paper, we provide a formal characterization of a generalized constraint satisfaction problem (CSP) that can be used to model many types of design/planning problems, and the architecture of an imlemented reasoning system for solving this problem. The architecture includes a truth maintenance system (TMS) which is specifically designed to reason about the relationships expressed in the constraints as a problem solution evolves. The CSP consists of two types of data. The first type of datum corresponds to assignments that are handled by the problem solver, and the second type corresponds to constraint terms handled by the TMS. The dependency network, representing the relationships among constraint terms, is static and generally quite small, depending on the number of constraint terms. Also, justifications are never manipulated (only evaluated). This results in an architecture that makes efficient use of both space and time. The need for efficient TMSs, even though these might deal only with certain classes of problems, is underscored by the fact that general purpose TMSs have often been found to be highly inefficient for solving large problems. We also show how certain instances of the generalized CSP can be formulated as an integer programming problem, special cases of which can be solved efficiently using mathematical (integer) programming techniques.

Most research on rule-based inference under uncertainty has focused on the normative validity and efficiency of various belief-update algorithms. In this paper we shift the attention to the inputs of these algorithms, namely, to the degrees of beliefs elicited from domain experts. Classical methods for eliciting continuous probability functions are of little use in a rule-based model, where propositions of interest are taken to be causally related and, typically, discrete, random variables. We take the position that the numerical encoding of degrees of belief in such propositions is somewhat analogous to the measurement of physical stimuli like brightness, weight, and distance. With that in mind, we base our elicitation techniques on statements regarding the relative likelihoods of various clues and hypotheses. We propose a formal procedure designed to (a) elicit such inputs in a credible manner, and, (b) transform them into the conditional probabilities and likelihood-ratios required by Bayesian inference systems.

The conceptual design of user interfaces focuses on the specification of the structure of the dialog, independent of any particular implementation approach. While there is common agreement with respect to the importance of this activity, adequate methods and tools to support it are generally unavailable. The Dialog Charts (DCs) yield high level dialog schemas that are abstract enough to support the conceptual design of dialog control structures. They combine dialog concepts with widely accepted design principles, in a uniform diagramming framework. Specifically, the DCs distinguish between the dialog parties, provide for hierarchical decomposition and enforce a structured control flow. A clear set of guiding principles for the conceptual design of dialogs has yet to emerge. In this paper we have elected to focus on the notions of descriptive power and usable power, as they apply to conceptual dialog modeling tools. The conceptual descriptive power of the DCs is informally examined by applying them in a varied set of examples and relating them to their lower level counterparts, namely implementation dialog models like augmented transition networks or context-free grammars. The usable power of the DCs has been examined empirically through a qualitative study of their actual use by system designers. The Dialog Chart models were found by dialog designers to be a useful conceptual design tool, which exhibit the essential attributes identified for conceptual models.

In order to promote more creative solutions to Information Systems (IS) design problems, this paper identifies four roles that images or diagrams can play in the IS development process. These roles are characteristics of the interaction between the image and its creator or viewer, rather than of the diagram itself. One of these roles in particular, the constructive role, can do much to support the generation of creative designs, to the benefit of both systems developers and their clients. The goal of constructive systems thinking is to enhance the creative solution of IS development problems, and it cannot be reduced to a specific, highly structured technique. We can, however, outline a general approach to building constructive images: 1. Create a set of candidate analogies, elaborate them, and evaluate the degree to which each guides design of the target system. 2. Evaluate how completely this working set of analogies in forms the important aspects of the target system, and create additional analogies to fill any major gaps. 3. Over the relevant scope of each analogy, research its structure and dynamic interactions. Use these as templates within which to model the target system. 4. Validate this design and its functional implications with the system client, adjusting it as required. 5. Complete the design by removing details specific to the analogous system and adding those relevant to the target system.

A new approach to the joint selection of primary and secondary routes in a network with unreliable components is presented. The mathematical model captures the changes in the operational characteristics of the network when it adapts to failures. Lagrangean relaxation and subgradient optimization techniques are used to obtain good heuristic solutions to the problem, as well as lower bounds to be used as benchmarks against which the quality of the solution is assessed. Results of numerical experiments are reported, and directions for further enhancements of the model are discussed.

Numerous software products claiming to improve work group productivity are making their way into the marketplace. How is a manager to decide whether or not to invest in these products? This paper proposes a suitable method of formulating the evaluation problem for work group application products. The issue of normal software evaluation is considered first. Then group work is described and contrasted with individual work activities. A two-level evaluation strategy consisting of broad coverage and detailed analysis is introduced. Detailed analysis consists of functional performance, administration, and fit applied across task, group and communications domains. Examples, drawn from some of the products demonstrated at this Symposium, are used to illustrate how this methodology may be applied.

This paper presents an empirical study of the strategic contributions of automated teller machines (ATMs) to improving a bank branch's local deposit market share at the expense of its competitors. By extending previous models of deposit market share in branch banking to incorporate ATM technology variables, we develop a tool to provide answers and insights on key questions involving the evaluation of second order strategic impacts of information technology (IT) which have not previously been measured in this context. Our results suggest that a bank's ATM network membership decision is crucial to its later success in enhancing deposit market share via deployment of ATMs. However, we found little evidence that branch ATMs provide any competitive leverage to increase a branch's local deposit share.

Information technology (IT) is essential to many businesses, but there are few guidelines for determining the adequate level of investment in IT. The purpose of this paper is to further understanding of the mechanism of IT investment. Previous studies on IT investment are briefly presented. The authors performed six-mini case studies of large companies in five different industries; these studies addressed the questions of how firms define IT and how they manage their investment in IT. Our goal was to formulate a model of the relationship between IT investment and organizational performance. We present the model and pose questions for investigating this important relationship more closely. Findings of interest relate to the definition of IT, the importance of political considerations, the concept of an industry-based threshold investment, the conversion effectiveness of IT investment, and the concept of productive capacity. The most important finding relates to the separation of different types of IT investment and their logical matching to particular performance measures.

In this paper, we report results of an investigation where thirty subjects were observed performing subject-based search in an online catalog system. The observations have revealed a range of misconceptions users have when performing subject-based search. We have developed a taxonomy that characterizes these misconceptions and hypotheses about the causes of the misconceptions. Directions for improving search performance are also suggested.

Temporally Oriented Databases (TODBs) are database systems in which both historical and current data are accessed and treated with full symmetry. The growing interest in such systems is manifested recently in a number of research efforts focusing on a wide set of issues, ranging from the study of abstract conceptual models to the practical implementation of working systems. Attempts to implement TODBs have so far been at best preliminary, characterized by an ad hoc flavor, or have had a very limited scope. This dissertation research is an attempt to design a general purpose relational Temporally-Oriented Database Management System (TDMS), and examine the feasibility of its implementation along current theoretical concepts. The users view data in a TDMS as a temporally oriented, three dimensional cube; this is, in fact, implemented as a two layered data structure. The implementation model interrelates the external user view with an underlying functional view of the data , and specifies on the translation between these layers. The major principle in the implementation is the differentiation of attributes according to their temporal variation . This research uses this concept as an implementation strategy of TDMSs, and assesses this approach for dealing with the following primary questions: efficient ways to store and retrieve data, the integrity constraints needed to maintain the database consistency and the definitions and implementations of temporal operations in such systems. Further validation of the model was achieved through the development of a TDMS prototype. The prototype was developed using INGRES commands embedded in PASCAL programs on VAX/VMS, and provides a test bed for further studies of temporally oriented information systems.

Belief updating schemes in artificial intelligence may be viewed as three dimensional languages, consisting of a syntax (e.g. probabilities or certainty factors), a calculus (e.g. Bayesian or CF combination rules), and a semantics (i.e. cognitive interpretations of competing formalisms). This paper studies the rational scope of those languages on the syntax and calculus grounds. In particular, the paper presents an endomorphism theorem which highlights the limitations imposed by the conditional independence assumptions implicit in the CF calculus. Implications of the theorem to the relationship between the CF and the Bayesian languages and the Dempster-Shafer theory of evidence are presented. The paper concludes with a discussion of some implications on rule-based knowledge engineering in uncertain domains.

Spreadsheet programs and microcomputers have revolutionized information processing in organizations. Users have adopted spreadsheets to solve problems and circumvent the long delays encountered in dealing with the traditional information services department. A significant number of serious errors have been reported through the misuse of spreadsheet technology. This paper discusses several different contexts for the development of spreadsheet models and presents structured design techniques for these models. The recommended approach to spreadsheet analysis and design encourages the use of a block structure format for the worksheet and introduces Spreadsheet Flow Diagrams as a systems design tool. The objective of this design approach is to reduce the probability and severity of spreadsheet errors, improve auditability and promote greater longevity for spreadsheet models.

Tuple selection predicates (such as CITY = "Brighton*) provide the basis for both the Select and Join operators in the relational model (RM). Several proposed extensions to RM can be seen as relaxing the First Normal Form constraint of simple values to allow function-valued attributes. In order to take advantage of these complex values the class of allowable tuple selection predicates likewise needs to be extended. We propose a class of general tuple selection predicates based upon a general model allowing function values of any type constructed from some basic set of primitive domain types, and we investigate how various recent proposals fit into this general framework.

Inventory Management has been widely discussed in the literature. Recently, the so called "Just in Time” method received extensive publicity and was claimed to be one of the major factors of the Japanese industrial success. This, in turn, promoted a large campaign in the rest of the industrialized world, to adopt and imitate the "Just in Time” (JIT) policy. Corporate and plant managers focused attention and set up goals as to reach as closely as possible the Japanese inventory levels. Quite often, adoption of JIT disregarded the totally different nature of the business their companies engaged in, relative to Japanese industry. This paper clarifies the differences between two different industrial models: The "Assembly Lines” model versus the Hi-Tech Job Shop "Science Based” model and prescribes the inventory strategy appropriate for each of those models. It is shown that a fully automated Assembly Line type factory requires a “Just in Time” (minimal holding costs) inventory strategy, while the Science Based type should follow a more elaborate “optimal Penalty” type of policy.

The purpose of this paper is to define and critique the use of contingency theory in the field of Management Information Systems (MIS). The existence of such a theory is demonstrated through a detailed review of the MIS literature. The development of contingency theory in MIS is compared to the development of Organization Theory. The developments in the two fields have been remarkably similar and the field of MIS can benefit from the experiences of organization theorists. We argue that since MIS is at an early stage of development, it is now repeating some of the unproductive assumptions and lines of development of contingency theory. The conclusion from this analysis is that the contingency theory implicit in MIS research is inadequate. Progress in the field has been hampered by the adoption of a naive meta-theory and a narrow research perspective. This has resulted in highly mixed empirical results, a premature quantification strategy, and ill-defined concepts of performance and fit. A series of recommendations for improving the theoretical basis of MIS are given. These recommendations include relaxing the assumptions that constitute the naive meta-theory of a contingency theory in MIS. A more subjectivist, less functional, less unreflexive and less deterministic approach is advocated. In addition, changes in research methodologies are recommended. An increased emphasis on training in case study methodologies, longitudinal research and ethnographic approaches is suggested.

Flexible representations are required in order to understand and generate expert behavior. While production rules with quantifiers can encode experiential knowledge, they often have assumptions implicit in them, making them brittle in problem scenarios where these assumptions do not hold. Qualitative models achieve flexibility by representing the domain entities and their interrelationships explicitly. However, in problem domains where assumptions underlying such models change periodically, it is necessary to be able to synthesize and maintain qualitative models in response to the changing assumptions. In this paper, we argue for a representation that contains partial model components that are synthesized into qualitative models containing entities and relationships relevant to the domain. The model components can be replaced and rearranged in response to changes in the task environment. We have found this "model constructor" to be useful in synthesizing models that explain and generate expert behavior, and have explored its ability to support decision-making in the problem domain of business resource planning, where reasoning is based on models that evolve in response to changing external conditions or internal policies.

This study examines the psychometric properties of the short form measure of user information satisfaction (UIS) proposed by Ives, Olson and Baroudi [1983]. Based on extensive testing the questionnaire appears to be a reasonably valid and reliable measure. A framework for how this measure can be used to detect and diagnose problems with user satisfaction is presented, and illustrated via two case studies. Finally, recommendations and suggestions are made regarding the future use of this and other measures of user information satisfaction.

Information Systems Working Papers Series

Fourteen experienced users of two commercial spreadsheet packages, Lotus 123 and Multiplan, performed four tasks - two of entering spreadsheets and two of modifying those same spreadsheets. Their actions were videotaped and analyzed for incidents of errors. Over 450 errors were made, the majority of them centered around the visual properties of the spreadsheet packages. A classification of the errors is presented with an analysis of the causes governing the production of the errors. A discussion of the choices in the design of the interface which facilitated the production of these errors is also presented.

Information Systems Working Papers Series

Information Systems Working Papers Series

The conceptual design of user interfaces focuses on the specification of the structure of the dialog, independent of any particular implementation approach. While there is common agreement with respect to the importance of this activity, adequate methods and tools to support it are generally unavailable. The Dialog Charts (DCs) presented in this paper address this problem -- they support the conceptual design of dialog control structures. The DCs combine visual modeling (i.e., diagraming) with widely accepted design principles and an explicit model of dialog structures. As no clear evaluation criteria exist in this evolving area of dialog design, the preliminary assessment of the DCs takes the form of contrasting them with representative alternative design tools based on Augmented Transition Networks or Backus-Naur Form grammars. The DCs overcome some of the problems that seem to limit the usefulness of comparable approaches. An empirical investigation of the usable power of the DCs is currently underway at New York University, and a summary of this research activity concludes the paper.

This paper presents a model of the implementation process for dedicated packages and describes a research project to test the model undertaken with the cooperation of a major computer vendor. Data were collected from 78 individuals in 18 firms using the package and from the package vendor. The results of the study offer some support for the model along with suggestions for package implementation for both the customer and package vendor.

Quality Control sampling plans are normally chosen by using industrial and military standards. These standards do not involve economic considerations, and usually fail to meet cost considerations. The Information Economic approach presented in this paper suggests an easy to use methodology which determines the optimal plan for a given situation of a quality control Common quality control attributes like, AQL, LTPD, Operating Characteristics Curves and Quality Control Plans are shown as special cases of Information Economic Models. Theorems involving dominance among various of Quality Control plans are proved. The Blackwell Theorem on the relationship "Generally More Informative” is modified to accommodate to the Quality Control case. The major results of the paper include new algorithm to optimize the expected utility of decision makers. The value of information in Quality Control systems is assessed, and cost/effective analysis is carried out.

Various techniques have been proposed to ensure the safe, concurrent execution of a set of database transactions. Locking protocols are the most prominent and widely used of these techniques, with two-phase locking and tree-locking being but two examples of these protocols. A locking protocol defines a general set of restrictions on the placement of lock and unlock steps within transactions. In this paper we show that it is possible to further increase the potential level of concurrency of a set of transactions, within the context of a specific locking protocol, by further restricting the placement of lock and unlock steps within each transaction. We also discuss a variation of the tree-locking protocol that allows transaction to be locked with respect to a dynamically changing set of tree structures. In addition we define and discuss the concept of a concurrency cost function for a locked transaction. This cost function measures the potential for conflict of a transaction with other transactions.

Negotiation support is an important aspect of multiperson decision support systems. Besides mechanisms for representing and evolving group joint problem representations, such DSS should also provide an environment in which decision makers are supported in developing, analyzing and reinforcing their individual negotiation position. Recognizing the diversity of research approaches to negotiation modeling in the literature, this paper synthesizes an integrated model from which a knowledge-based individual negotiation support environment using tools from different areas can be designed. Role and architecture of such a component are described in the context of MEDIATOR, a database-centered negotiation support system under development at NYU.

Critical to the design of an historical database model is the representation of the “existence” of objects across the temporal dimension -- for example, the "birth," "death," or "rebirth" of an individual, or the establishment or dis-establishment of a relationship. The notion of the "lifespan" of a database object is proposed as a simple framework for expressing these concepts. An object's lifespan is simply those periods of time during which the database models the properties of that object. In this paper we propose the historical relational data model (HRDM) and algebra that is based upon lifespans and that views the values of all attributes as functions from time points to simple domains. The model that we obtain is a consistent extension of the relational data model, and provides a simple mechanism for providing both time-varying data and time-varying schemes.

This paper presents a framework for the implementation of strategic information systems. The framework draws on past research on implementation and takes into account the unique circumstances of strategic applications. The framework is illustrated with a case study of a money-center bank's cash management system based on a microcomputer. The microcomputer provides a powerful front end to the bank's traditional transactions processing system for a corporate treasurer who is a client of the bank. The impact of the system appears to be positive and the framework offers one model for viewing the implementation of strategic systems.

Information Systems Working Papers Series

The largescale ineffectiveness of current systems development methodologies may be attributed to the inaccuracy or inadequacy of their underlying assumptions concerning the systems development process. In this paper, we propose a descriptive, alternative model of the Information Systems (IS) design process. This model emphasizes the importance of constraints in defining the feasible design space, and of assumptions as a vehicle for discovering constraints. Moreover, rather than assuming that design activities occur in a logical and prescribed sequence, as the current dominant model, the Systems Development Life Cycle (SDLC) does, the Constraints/Assumptions (C/A) Model focuses on the interdependent nature of design activities. The importance of developing and validating alternative models of the system design process is evident from three sources. First, there is the paucity of empirical research on systems design, which we attribute to a scarcity of theory to guide such research. Second, educators evince serious doubt as to our ability to educate students in this process. Third, the widespread inability of professional systems designers to develop systems on schedule, within budget, and providing the full set of specified functions is disconcerting, if not appalling. Previous research suggests that superior designs are produced when both clients and designers regard the IS design process as a learning experience, and work to educate each other. The Constraints/Assurnptions Model further elaborates this mutual learning thesis, by differentiating what clients learn from designers and what designers learn from clients. The C/A Model asserts that, at any stage in the design process, two dialogues occur simultaneously. The client/designer dialogue elaborates the design space, i.e., a set of constraints on the design process specifying required performance and function, the organizational and political climate, the resources available for developing and operating the system, etc. The designer/team dialogue, on the other hand, focuses on the generation of a working solution to the design problem, its validation with respect to technical feasibility and its congruence with the acceptable design space, and its elaboration into an implementable design. Both the design space and the working design are inputs to each dialogue, and their interdependence results from each dialogue's ability to modify only its own product.

Information Systems Working Papers Series

This paper describes the philosophy underlying the development of an intelligent system to assist in the formulation of large linear programs. The LPFORM system allows users to state their problem using a graphical rather than an algebraic representation. A major objective of the system is to automate the bookkeeping involved in the development of large systems. It has expertise related to the structure of many of the common forms of linear programs (e.g. transportation, product-mix and blending problems) and of how these prototypes may be combined into more complex systems. Our approach involves characterizing the common forms of LP problems according to whether they are transformations in place, time or form. We show how LPFORM uses knowledge about the structure and meaning of linear programs to construct a correct tableau. Using the symbolic capabilities of artificial intelligence languages, we can manipulate and analyze some properties of the LP prior to actually generating a matrix.

Information Systems Working Papers Series

Information processing technology has had a profound impact on organizations since the widespread adoption of computers in the 1950s. How will this technology influence management and organizations in the 1990s? Will there be a revolution of management theory and practice? The purpose of this paper is to present 1) significant trends in the technology and 2) to discuss how senior management can take advantage of the technology to change the fundamental nature of a business. What does the senior manager have to do to see that information technology is employed effectively in the 1990s? What technology is most appropriate for the firm and how should it be selected? How does the firm implement new technology?

The completion of a typical project hinges upon receiving all the purchased components by the time they are scheduled to be used. Some of these components may have long stochastic lead times, so the project manager is tempted to order them ASAP, to avoid the (usual) high penalties associated with delays. (The penalties may be tangible or intangible, but we assume that they can be measured by monetary units.) This in turn may bring about excessive inventory holding costs. Clearly an optimization is called for to minimize the total expected cost of the project. This is achieved by timing the orders optimally, for the one component case as well as for the n component general case.

This study reviews 57 MIS articles employing statistical inference testing published in leading MIS journals over the last five years. The statistical power of the articles was evaluated and found on average to fall substantially below the accepted norms. The consequence of low power is that it can lead to misinterpretation of data and results. Collectively these misinterpretations result in a body of MIS research that is built on potentially erroneous conclusions.

Competitive Information Systems (CIS) are information systems which help a company to obtain and sustain a competitive edge. Before any such CIS can be implemented, the idea for it has to be formulated. The paper describes a way to systematically stimulate ideas by asking questions. It discusses the question generating mechanism as well as ways to focuses these questions. It shows an implemented DSS, which aids the described process and contains inference mechanisms of expert systems. This DSS uses a symbiotic approach between system and user.

Research on expert-novice differences falls into two complementary classes. The first assumes that novice skills are a subset of those of the expert, represented by the same vocabulary of concepts. The second approach emphasizes novices' misconceptions and the different meanings they tend to attribute to concepts. Our evidence, based on observations of problem solving behavior of experts and novices in the area of mathematical programming, reveals both type of differences: while novices are to some extent underdeveloped experts, they also attribute different meanings to concepts. The research suggests that experts' concepts can be characterized as being more differentiated than those of novices, where the differentiation enables experts to categorize problem descriptions accurately into standard archetypes and facilitates attribution of correct meanings to problem features. Our results are based on twenty-five protocols obtained from experts and novices attempting to structure problem descriptions into mathematical programming models. We have developed a model of knowledge in the LP domain that accommodates a continuum of expertise ranging from that of the expert who has a highly specialized vocabulary of LP concepts to that of a novice whose vocabulary might be limited to high school algebra. We discuss the normative implications of this model for pedagogical strategies employed by instructors, textbooks and intelligent tutoring systems.

This paper reports on a recent survey of 25 of the largest PC users in the financial industry. The survey focused on non-data processing use of PC's in end user divisions. Descriptive data on the uses, users, quality control, and decision making practices are examined. A stage theory, based on PC telecommunication configurations, is proposed to account for organizational differences in utilization and management.

The paper discusses the value of information when a number of independent sources provide information related to a common set of states of nature. The starting point is the Information Economic model of Information Structures. The model is augmented to represent independence of informational sources by means of orthogonality of the information structures. A new mathematical operator, orthogonal product, is defined and its properties are probed. It is shown that this operator maintains some mathematical properties such as closure, association, unity element, null element, etc. It is demonstrated how the orthogonal product represents the notion of multi-source information. The paper proves that an orthogonal product is generally more informative than its multipliers, namely, if cost is not considered a constraining factor, then there is a nonnegative value to obtaining a second opinion. The paper concludes with a numerical example and a discussion on the applicability of the model of orthogonality.

It is argued that the consistent evidence of psychosomatic stress problems associated with the use of computer systems in offices may be the result of the redesign of work stemming from the implementation of these systems. This proposition is tested with samples of mortgage loan servicing clerks (N=1282) and financial investment officers (N=131) in 78 Mutual Savings Banks. Evidence is presented that workers, in the structured job, making more intense use of computer systems do have a poorer task environment than those making less use of systems. This occurs because the resulting job tends to be more interdependent and involve greater demands than without an application system, although several other factors interact reducing these affects somewhat. Productivity is also increased, and the task environment interacts in complex ways with the characteristics of the system. Various strategies are suggested for correcting the negative effects including providing workers with more decision latitude in the selection of work methods and making it easier to obtain problem solving assistance. No relationship is found between productivity, performance, or the task environment and system use or characteristics for the semi-structured job. It is concluded that although application systems have considerable negative consequences for office work environments, the strength of these affects depend upon the level of the job.

Information Systems Working Papers Series

The debate regarding the impact of technology on work organization and the quality of work life is gaining momentum as a result of office automation and its influence on large numbers of "office” jobs. This paper argues that current approaches to implementation of office automation technology are based on faulty assumptions about both technology and work organization. The result has been systems which fail to live up to the claims of dramatic increases in productivity that typically accompany their implementation. The paper compares information technology to production technology and reviews relevant research on impacts of technology on work organization and management control. An approach to the implementation of office automation systems is proposed, which is based on a new concept of "offices" and "office work”. In essence, technology is the facilitator of a wider range of choices in organization and control of office work than was previously possible. When these new choices are considered, the technology can be applied to both improve organizational productivity and effectiveness and to accommodate employees’ needs for flexibility and a high quality work life.

This paper considers four aspects of end user computing: false stereotypes, information tasks supported, inconsistency between flexibility and formal systems, and contributions made by information technology.

The maintenance of large information systems involves continuous design modifications to designs in response to evolving business conditions or changing user requirements. Because of the complexity barrier associated with engineering such systems, changes can be ad hoc and prone to errors. Based on our observations of such a process in the oil industry, we believe that the systems maintenance activity would benefit greatly if the process knowledge reflecting the teleology of a design could be captured and used in order to reason about changing requirements, and to design parts of systems that might be “similar” to existing ones. In this paper, we describe a partially implemented formalism called REMAP (REpresentation and MAintenance of Process knowledge) that accumulates design process knowledge to manage systems evolution. To accomplish this, REMAP acquires and maintains dependencies among the design decisions made during a prototyping process as well as the general domain-specific design rules on which such dependencies are based. This knowledge can then be applied to prototype refinement, systems maintenance, and the re-use of existing designs to construct “similar” design fragments.

Over the last decade there have been several efforts at building knowledge based "expert systems”, mostly in the scientific and medical arenas. Despite the fact that almost all such systems are in their experimental stages, designers are optimistic about their eventual success. In the last few years, there have been many references to the possibility of expert systems in the management literature. However, what is lacking is a clear theoretical perspective on how various management problems differ in nature from problems in other domains, and the implications of these differences for knowledge based decision support systems for management. In this paper, I examine some of these differences, what they suggest in terms of the functionality that a computer based system must have in order to support organizational decision making, and the scope of such a system as a decision aid. The discussion is grounded in the context of a computer based system called PLANET that exhibits some of the desired functionality.

This paper describes a mathematical programming generator that interprets problem statements written in the algebraic notation found in journal articles and text-books and outputs statements in the 'MPS format’ used by IBM’s MPSX mathematical programming system. The system has been implemented in the APL programming language. Although originally designed for stand-alone use, it is currently being used as a component in an expert system that will help users formulate large linear programming models, The paper describes the syntax of the problem definition language and gives some illustrative examples. There are several unique features. First, the user can define objective function, constraint and right-hand-side coefficients as APL expressions. This leads to concise problem statements and also reduces data storage and processing requirements. Second, the system supports an integrated data base query language. Finally, there are a number of aids for model maintenance and sensitivity analysis. The last section of the paper describes the use of MPGEN in the expert system context.

A number of DSS for supporting decisions by more than one person have been proposed. These can be categorized by spatial distance (local vs. remote), temporal distance (meeting vs. mailing), commonality of goals (cooperation vs. bargaining), and control (democratic vs. hierarchical). Existing frameworks for model management in single-user DSS seem insufficient for such systems. This paper views multiperson DSS as a loosely coupled system of model and data bases which may be human (the DSS builders and users) or computerized. The systems components have different knowledge bases and may have different interests. Their interaction is characterized by knowledge sharing for uncertainty reduction and cooperative problem-solving, and negotiation for view integration, consensus-seeking, and compromise. Requirements for the different types of multiperson DSS can be formalized as application-level communications protocols. Based on a literature review and recent experience with a number of multiperson DSS prototypes, artificial intelligence-based message-passing protocols are compared with database-centered approaches and model-based techniques, such as multicriteria decision making.

This paper describes a problem solver called PLANET that has been developed in collaboration with a large computer manufacturing company to assist planning managers with the formulation and maintenance of planning models for resource allocation. PLANET is equipped with the primitives that enable it to preserve much of the richness of the process of the planning activity, namely, the generation of symbolic alternatives, and for the expression of domain specific knowledge which enables it to synthesize these alternatives into an overall planning model. This knowledge is maintained in a “meta-model.” In contrast to modeling systems which allow for parametric perturbations of an algebraic model, PLANET's meta-model provides it with the capability for systematic variations in the symbolic model assumptions, with concomitant structural variations induced in the algebraic model that reflect the interdependencies of those assumptions. Whenever previously held assumptions change, PLANET uses the existing model as a point of departure in formulating the revised plan. In this way, the program is able to take cognizance of the ongoing nature of organizational problem solving, and can serve an important decision support function in maintaining and reasoning about evolving plans.

An earlier study (Lucas, 1984), while finding support for the Hickson (1971) framework for organizational power, found that department managers in three major industrial firms rated their information services department as the least critical of five departments, for success in their industry. This study polled 37 plant managers from the same firms. Plant managers, like their department manager counterparts, considered the information services department least critical for success. Centralization of decision making was found to be positively related to information services department power. Finally, implications of the study on distributed versus centralized information processing are discussed.

This paper reports the results of an exploratory study that investigated expert and novice debugging processes with the aim of contributing to a general theory of programming expertise. The method used was verbal protocol analysis. Data was collected from sixteen programmers employed by the same organization. First, an expert-novice classification of subjects was derived from information based on subjects’ problem solving processes; the criterion of expertise was the subjects' ability to effectively chunk the program they were required to debug. Then, significant differences in subjects’ approaches to debugging were used to characterize programmers' debugging strategies. Comparisons of these strategies with the expert-novice classification showed programmer expertise based on chunking ability to be strongly related to debugging strategy. The following strategic propositions were identified for further testing: 1. (a) Experts use breadth-first approaches to debugging and, at the same time, adopt a system view of the problem area. (b) Experts are proficient at chunking programs and hence display smooth-flowing approaches to debugging. 2. (a) Novices use breadth-first approaches to debugging but are deficient in their ability to think in system terms. (b) Novices use depth-first approaches to debugging. (c) Novices are less proficient at chunking programs and hence display erratic approaches to debugging.

It is accepted, generally, that Information Systems (IS) personnel have specific needs and present a unique set of problems for management. This position is challenged. It is our contention that IS personnel exhibit relatively few differences when compared with other, similar, occupational groupings. This does not imply, however, that the unique aspects of IS work are unimportant, or that attention need not be focused on understanding the determinants of particular behavioral outcomes. Rather, it suggests, that the effective management of the IS human resource may well rest more on general management practices than is commonly believed. This paper identifies managerial complaints about IS personnel and provides a summary of research findings. A model, based on accepted principles of worker behavior, intended to integrate much of the previous work, is presented. The IS staff is classified into categories and propositions about each are developed. Specific research questions are formulated which, when answered, will provide more insight into the unique aspects of IS work and will identify managerial strategies for improving employee performance and well being.

In search of the appropriate semantics for the inclusion of structures and operations that will meet the needs of a wide class of users interested in a database system supporting temporal views of their data, the paper includes a discussion of many problems that must be addressed. Salient features of the author’s Historical Relational Database Model (HRDBM) are presented, and some subtle nuances that time brings to the development of an historical relational algebra are illustrated. Along the way, a number of observations and guidelines are presented that may help guide the search for an historically relationally complete database model and query languages.

Current DSS research is rather fragmentary, and typically myopic--it centers either on the decision situation which DSS support, or on DSS tools or generators. In this paper we adopt a comprehensive view of DSS emphasizing their systemic nature. This entails identifying the links among the five aspects that classically characterize a system: 1, the environment, i.e., decision situations and access patterns; 2. the function (within this environment), i.e., types and levels of decision support; 3. the functional components that make it up, i.e., dialog, data, and model management; 4. the arrangement, i.e., the linkages among the components and the assignment of functions to modules; and 5. the resources consumed, i.e., hardware, software, human skills, and data. The systemic view provides a concrete framework for the effective design of DSS, and serves as a basis for accumulating DSS research results.

Information Systems Working Papers Series

If general principles for the design of computer mediated work are to emerge, the interplay between the work environment and the characteristics of application systems must be understood better. The results of a study of 620 Claims Representatives in the Social Security Administration, whose job differed only in the form of the application system interface used, are reported. Interactions with clients, operators’ perceived task environment and well being are all seen to be influenced by the type of system interface used. Mental strain symptoms is shown to be an important indicator of operator well being. Implications for designers, managers and researchers are considered.

Information Systems Working Papers Series

Semantic query simplification utilizes integrity constraints enforced in a database system for reducing the number of tuple variables and terms in a relational calculus query. To a large degree, this can be done by a system that is external to the DBMS. The paper advocates the application of database theory in such a system and describes a working prototype of an external semantic query simplifier implemented in Prolog. The system employs a graph-theoretic approach to integrate tableau techniques and algorithms for the syntactic simplification of queries containing inequality conditions. The use of integrity constraints is shown not only to improve efficiency but also to permit more meaningful error messages to be generated, particularly in the case of an empty query result. The paper concludes with outlining an extension to the multi-user case.

Query processing in databases can be divided into two steps: selecting an 'optimal' evaluation strategy, and executing it. We first present elementary nested loop and relational algebra algorithms for query execution and point out some opportunities for improving their performance. A survey of optimization strategies, structured in query transformation techniques and access planning methods, follows. Finally, extensions for special-purpose query systems are briefly addressed.

The simultaneous optimization of multiple queries submitted to a database system may lead to substantial savings over the current approach of optimizing each query separately. Isolating common subexpressions in multiple queries and treating their execution as a sharable resource are important prerequisites. This chapter presents techniques for recognizing, supporting, and exploiting common subexpressions in record-oriented, relational algebra, domain relational calculus, and tuple relational calculus query representations. It also investigates preconditions that transaction management mechanisms must satisfy to make multiple query optimization effective.

In many decision support systems, multiple decision methods and models must be combined for solving a complex problem. Expertise is required for selecting, adapting and coordinating appropriate models. This paper describes the design and implementation of a knowledge-based model management system called the Actuarial Consulting System (ACS). The ACS supports actuaries in making pricing decisions in the domain of life insurance. Actuarial knowledge is organized using a graph formalism called Formula Derivation Network (FDN), represented in Prolog as a hierarchy of predicates. On the user level, a Problem Analyzer converts a problem specification by the user into a search problem on the stored collection of FDNs. Using different search strategies, including human expert rules, the Surface Planner generates an efficient solution strategy (sequence of models). At the lowest level, a Plan Executor retrieves or requests model data and issues appropriate function calls to a subroutine library.

This paper focuses on data management aspects of computerized decision support systems which use interactive multiple criteria decision methods. In this context, we point out the technical requirements for such systems and the importance of the data management tool to MCDSS. After a discussion of candidate data models (i.e. relational, hierarchical, and network), we examine the criteria to use in choosing the data model for MCDSS. In the last part of this paper, we review some database management services which support data definition, data manipulation, and data integrity within the multiple-criteria decision making framework. These services guide us when designing the appropriate architecture for the MCDSS’s data component.

An optimizing translation mechanism for the dynamic interaction between a logic-based expert system written in PROLOG and a relational database accessible through SQL is presented. The mechanism makes use of an intermediate language that decomposes the optimization problem and makes the proposed approach target-language independent. It can either facilitate expert system - database interaction, e.g., when integrating expert systems into business systems, or augment existing database with (external) deductive capabilities.

Although a large number of natural language database interfaces have been developed, there have been few empirical studies of their practical usefulness. This paper presents the design and results of a field evaluation of a natural language system - NLS - used for data retrieval . A balanced, multifactorial design comparing NLS with a reference retrieval language, SQL, is described. The data are analyzed on two levels: work task (n=87) and query (n=1081). SQL performed better than NLS on a variety of measures, but NLS required less effort to use. Subjects performed much poorer than expected based on the results of laboratory studies. This finding is attributed to the complexity of the field setting and to optimism in grading laboratory experiments. The methodology developed for studying computer languages in real work settings was successful in consistently measuring differences in treatments over a variety of conditions.

Information Systems Working Papers Series

Information Systems Working Papers Series

Are natural language systems for database queries meeting their goals? And, are these goals appropriate? The recently completed Advanced Language Project at New York University combined a field experiment with two laboratory studies to examine these issues by comparing performance between subjects using the formal database language SQL and subjects using the prototype natural language system, USL. This paper describes the design and results of the larger laboratory experiment. The results presented offer some promise for the usability of natural language under certain conditions.

In this paper, we propose a new approach to the analysis and design of management information systems. While previous methods were either syntactic in nature (structured analysis and design, problem statement languages) or focused on the user-analyst interaction (user-controlled design, prototyping), our method - while compatible with both - additionally incorporates the experiential knowledge gained from the thousands of systems in operation today. This goal is achieved through providing the systems analyst with a business systems architecture (BSA) consisting of two parts: a set of rules representing the knowledge about a generalized application domain, and a domain-specific database architecture that enforces these rules. The rules can also be used to check the design of existing systems and to guide the analysis process for new systems. The paper describes a BSA for transaction processing systems. The approach can be applied in a similar way to other application domains such as decision support systems.

The central hypothesis of this paper is that database design and systems design in general can be simplified considerably by tailoring the design methods to a suitable range of applications. Domain-specific knowledge can be incorporated into a specialized database architecture that leaves the designer with the task to specify only the application-specific parts. Based on an analysis of business constraints, we propose such an architecture for the domain of business transaction processing. The architecture offers several data and transaction management services, special-purpose sub-databases, and design checking rules to be used by the application designer. Two services, input management and audit and control services, are described in more detail.

Database queries explicitly containing existential and universal quantification become increasingly important in a number of areas such as integrity checking, interaction of databases, and statistical databases. Using a concept of range nesting in relational calculus expressions, the paper describes evaluation algorithms and transformation methods for an important class of quantified relational calculus queries called perfect expressions. This class includes well-known classes of "easy" queries such as tree queries (with free and existentially quantified variables only), and complacent (disconnected) queries.

Information Systems Working Papers Series

Information Systems Working Papers Series

In this paper we survey some issues relating to the language interfaces provided by DSS. We do this from the point-of-view of the designer of generalized software for building DSS. We first describe a fairly general architecture for such software. This is followed by a brief examination of the language interface components. We list the functions that must be provided through the language facilities of a DSS together with a number of implementation issues such as User-friendliness, degree of procedurality and interpretation versus compilation. This provides a useful framework for the comparative evaluation of DSS packages. Finally we discuss some possible directions for future development including specialized formal languages and the potential of English-like 'semi-natural' languages.

Much of the gap between the potential of Information Systems and their realization can be explained by behavioral and organizational issues. This paper outlines the scope of behavioral research in information systems using selected examples, identifies several principles underlying this work, and suggests reasons why the results of this research are of value.

The problem of managing an intercontinental container transportation system is used as an example of how knowledge from the areas of database design, management science, and human factors research can be combined in the design of a decision support system. Using a new representation of time-related database objects, we first present a logical data model of a container transportation system. A hierarchically distributed decision support system can be based on this model. A physical database structure is proposed and a survey of partial optimization models to be used in the decision support system is given.

The concept of an historical database is introduced as a tool for modelling the dynamic nature of some part of the real world. Just as first-order logic has been shown to be a useful formalism for expressing and understanding the underlying semantics of the relational database model, intensional logic is presented as an analogous formalism for expressing and understanding the temporal semantics involved in an historical database. The various components of the relational model, as extended to include historical relations, are discussed in terms of the model theory for the logic ILs, a variation of the logic IL formulated by Richard Montague. The modal concepts of intensional and extensional data constraints and queries are introduced and contrasted. Finally, the potential application of these ideas to the problem of Natural Language Database Querying is discussed.

Predictions of the effects of office automation on organizations vary widely. This article focuses on changes in individual work patterns, management control, and organizational structure that may occur as a result of implementation of office technology. The most significant change predicted is that organizations will no longer be limited by a central office work environment operating between the traditional office work hours of nine and five. Computer and communications technology will facilitate the relaxing of those physical constraints as necessitated by social and economic pressures. Relevant research to date regarding the effects of the new technology on organizational behavior is reviewed. Management guidelines for preparing for the coming changes are included.

Information Systems Working Papers Series

This paper critically reviews measures of user information satisfaction and selects one for replication and extension. A survey of production managers is used to provide additional support for the instrument, eliminate scales that are psychometrically unsound, and develop a standard short form for use when only an overall assessment of information satisfaction is required and survey time is limited.

This study examines the impact of on-line system response time on CRT operator productivity and job satisfaction. It was predicted that increase in response time would affect total transaction volume and total errors adversely, that is, total transaction volume would decrease with longer response times and total errors would increase. Total productive transactions, the difference between total transactions and total errors, was expected to decrease as response time increased. Operator job satisfaction was also expected to decrease. The study confirmed the prediction with regard to total transactions and productive transactions: both decreased as response time increased. Total errors actually decreased as response time increased, up to times of 12 seconds. When response time exceeded 12 seconds, errors increased. The impact of response time on productivity suggests nearly all transactions should be completed in 12 seconds or less. Beyond this level, the organization in the study suffered severe penalties in lost productivity. A relationship was also found between increased response times and reduced job satisfaction.

An approach to measuring information systems properties is developed and tested with data from 38 systems in the same application area (mortgage loan servicing). The results provide support for the notion that general measures of system properties can be made operational and that these measures may be useful to other researchers.

Information Systems Working Papers Series

Information Systems Working Papers Series

Computer chargeback systems are installed to meet various data processing objectives. One objective is to increase user involvement in decisions regarding information systems development and use. Presumably, increasing user involvement will result in more effective information systems. In a field study we examine the relationship between various characteristics of a computer chargeback system, the quality of the chargeback system's user interface, user involvement and user attitudes about information systems services. Suggestions are given both for the practicing information systems manager and for future information systems researchers.

Decision Support Systems (DSS) represent an ever increasing portion of the investment in computer-based systems in organizations. Unlike earlier systems which aimed to replace existing clerical processes with faster, more efficient clerical processes, DSS attempt to extend and expand the capabilities of organizational decision makers. This fundamental difference in purpose between DSS and clerical systems causes our existing notions about system success to be inadequate. This paper explores the issue of DSS success, asking what it is, how it can be measured, and what can be done to facilitate it.

Data translation and transaction translation are two major problems that have to be solved in order to achieve the coexistence of heterogeneous distributed databases. In this paper we discuss the problem of transaction translation. The nature of the problem is explored by developing direct translations of transactions between the relational and hierarchical and network models. Methods for mapping a hierarchical or network schema to an equivalent relational schema are presented. The relational operators projection, selection, join, insertion. deletion and update are translated to equivalent hierarchical and network operations.

Considerable prescriptive literature exists which advocates user involvement in the development of information systems and suggests alternative mechanisms by which such involvement can be increased. However, formal empirical studies investigating user involvement are few in number, fragmented, and generally methodologically flawed. Furthermore, they do not provide the strong support for user involvement that the prescriptive literature would lead one to expect. This paper critically examines past studies of user involvement, focusing on methodological and measurement issues. The relationships between user involvement and system quality, system usage, information satisfaction, and user attitudes are considered. Suggestions for future research are discussed.

Systems supporting new products or services and driven by organizationally dictated deadlines limit user input and planning prior to design. An on-line model of implementation is proposed calling for constant re-evaluation and re-direction of the implementation as these shifting projects develop. An application of the model with a 200,000 line government systems is described.

This paper presents the results of a exploratory study of problem solving by top management in a sample of firms. The purpose of the research is to determine the nature of the information needed by top management and the role of computer-based system in suppling it. The paper classifies problems according to their nature and explores the information sources used to solve the problems. It is clear that computer-based information can aid in the solution of some of these top management problems, but probably through unconventional systems. The implications of the study for systems design are discussed.

Information Systems Working Papers Series

This paper develops the design of a data base system to support operations research models in the context of an integrated planning system involving a number of different users and computer programs. The requirements for such a system are described, a “network” data base schema is developed and the schema and command language are illustrated through a specific example.