Knowledge that Transforms

We examine the verification of large knowledge-based systems. When knowledge bases are large, the verification process poses several problems that are usually not significant for small systems. We focus on decompositions that allow verification of such systems to be performed in a modular fashion. We identify a graphical framework, that we call an ordered polytree, for decomposing systems in a manner that enables modular verification. We also determine the nature of information that needs to be available for performing local checks to ensure accurate detection of anomalies. We illustrate the modular verification process using examples, and provide a formal proof of its accuracy. Next, we discuss a meta-verification procedure that enables us to check if decompositions under consideration do indeed satisfy the requirements for an ordered polytree structure. Finally, we show how the modular verification algorithm leads to considerable improvements in the computational effort required for verification as compared to the traditional approach.

The inherent uncertainty pervasive over the real world often forces business decisions to be made using uncertain data. The conventional relational model does not have the ability to handle uncertain data. In recent years, several approaches have been proposed in the literature for representing uncertain data by extending the relational model, primarily using probability theory. The aspect of database modification, however, has not been addressed in prior research. It is clear that any modification of existing probabilistic data, based on new information, amounts to the revision of one's belief about real-world objects. In this paper, we examine the aspect of belief revision and develop a generalized algorithm that can be used for the modification of existing data in a probabilistic relational database. The belief revision scheme is shown to be closed, consistent, and complete.

In this paper we first present an empirical study of groupware use illustrating problems that users faced with restricted feedback about others' activities. Awareness can aid users in learning interdependencies, and in forming conventions to regulate system use and information-sharing. As a solution to providing awareness, we integrate the framework of organizational memory with intelligent agent technology to provide a coordination mechanism that enables the structuring of awareness events and gives information about the users' feedback control. In the proposed model, feedback control relationships are captured into a multilayered model of organizational memory and transferred to users by agents-facilitators. The approach is based on a system dynamics approach to organizational learning.

We propose priority pricing as an on-line adaptive resource scheduling mechanism to manage real-time databases within organizations. These databases provide timely information for delay sensitive users. The proposed approach allows diverse users to optimize their own objectives while collectively maximizing organizational benefits. We rely on economic principles to derive priority prices by modeling the fixed-capacity real-time database environment as an economic system. Each priority is associated with a price and a delay, and the price is the premium (congestion toll resulting from negative externalities) for accessing the database. At optimality, the prices are equal to the aggregate delay cost imposed on all other users of the database. These priority prices are used to control admission and to schedule user jobs in the database system. The database monitors the arrival processes and the state of the system, and incrementally adjusts the prices to regulate the flow. Because our model ignores the operational intricacies of the real-time databases (e.g., intermediate queues at the CPU and disks, memory size, etc.) to maintain analytical tractability, we evaluate the performance of our pricing approach through simulation. We evaluate the database performance using both the traditional real-time database performance metrics (e.g., the number of jobs serviced on time, average tardiness) and the economic benefits (e.g., benefits to the organization). The simulation results, under various database workload parameters, show that our priority pricing mechanism not only maximizes organizational benefits but also outperforms in all aspects of traditional performance measures compared to frequently used database scheduling techniques, such as first-come-first-served, earliest deadline first and least slack first.

Notably absent in previous research on inductive expert systems is the study of meanrisk trade-offs. Such trade-offs may be significant when there are asymmetries such as unequal classification costs, and uncertainties in classification and information acquisition costs. The objective of this research is to developmodels to evaluate mean-risk trade-offs in value-based inductive approaches. We develop a combined mean-risk measure and incorporate it into the Risk-Based induction algorithm. The mean-risk measure has desirable theoretical properties (consistency and separability) and is supported by empirical results on decision making under risk. Simulation results using the Risk-Based algorithm demonstrate: (i) an order of magnitude performance difference between mean-based and risk-based algorithms and (ii) an increase in the performance difference between these algorithms as either risk aversion, uncertainty, or asymmetry increases given modest thresholds of the other two factors.

Error search and correction are major contributors to software development cost, yet typically uncover only a small fraction of software errors. Postrelease errors, i.e., those that are only observed after a system is released, threaten a variety of potential failures and consequences, each with low individual probability of occurrence. The combined effect of postrelease errors can and often does result in a significant rate of occurrence of these potential failures, with unpredictable consequences and severity. One particular source of postrelease errors that has received extensive publicity is the year 2000, or Y2K, error. The modeling in this research report suggests that testing probably needs to be conducted over more than half of the useful life of a system in order to discover even one-third of the total errors in the system. It suggests that short product lifecycles, lifetime testing, and effective feedback loops for error reporting are necessary to assure reliable software.

The most difficult challenge facing a market leader is maintaining its leading position. This is especially true in information technology and telecommunications industries, where multiple product generations and rapid technological evolution continually test the ability of the incumbent to stay ahead of potential entrants. In these industries, an incumbent often protects its position by launching prematurely to retain its leadership. Entry, however, happens relatively frequently. We identify conditions under which an entrant will launch a next generation product thereby preventing the incumbent from employing a protection strategy. We define a capabilities advantage as the ability to develop and launch a next generation product at a lower cost than a competitor, and a product with a greater market response is one with greater profit flows. Using these definitions, we find that an incumbent with a capabilities advantage in one next generation product can be overtaken by an entrant with a capabilities advantage in another next generation product only if the entrant's capabilities advantage is in a disruptive technology that yields a product with a greater market response. This can occur even though both next generation products are available to both firms. We also show that the competition may require the launching firm to lose money at the margin on the next generation product.

Previous research has demonstrated that the production of information services can be characterized at the aggregate economy-wide level by the Cobb-Douglas production function. However, the underlying production process at the firm level has not yet been ascertained. The objective of this paper is to determine the form of the production process for information systems services at the firm level by conducting an empirical analysis of IS budget data. The production of information services is modeled using a production function with two inputs, hardware and personnel. We estimate various econometric specifications to determine several characteristics of the provision of information services, including the allocation of the information systems budget to its two largest components—hardware and personnel—and its implications for the form of the production function. After controlling for industry sector, we find that the ratio of personnel to hardware is independent of scale, which indicates a homothetic production function. We also find that the ratio of factor shares is constant with time, consistent with the Cobb-Douglas production function.We conclude that the underlying form of the production function is the same at the level of both the firm and the economy. Our analysis demonstrates how the application of production theory to the production of information services can yield useful insights from both a theoretical and managerial perspective.

For more than a decade NEC dominated the Japanese PC market with its PC-98 architecture, which was incompatible both with its major Japanese rivals and the global PC standard. However, NEC was powerless to prevent the introduction of Japanese versions of Windows 3.1 and 95 that ran on its competitors' architectures as well as on the PC-98, unifying the Japanese PC market and creating a common set of application programming interfaces for all Intel-based Japanese PCs. The introduction of Windows rendered obsolete the large DOS-based software library that had provided strong positive externalities for the NEC architecture. Absent those advantages, the market share of the PC-98 standard fell from 60% to 33% in five years, and NEC finally abandoned the PC-98 in favor of the global standard. An examination of the unusual rise and fall of the PC-98 shows how victory in a standards competition can be negated by the introduction of a new architectural layer that spans two or more previously incompatible architectures.