Showing papers in "Expert Systems With Applications in 1995"

Cases, scripts, and information-seeking strategies: On the design of interactive information retrieval systems

Abstract: The support of effective interaction of the user with the other components of the system is a central problem for information retrieval. In this paper, we present a theory of such interactions taking place within a space of information-seeking strategies, and discuss how such a concept can be used to design for effective interaction. In particular, we propose a model of information retrieval system design based on the ideas of: a multidimensional space of information-seeking strategies; dialogue structures for information seeking; cases of specific information-seeking dialogues; anti, scripts as distinguished prototypical cases. We demonstrate the use of this model by discussing in some detail the MERIT system, a prototype information retrieval system, that incorporates these design principles.

Effectiveness of neural network types for prediction of business failure

Abstract: The study examines the effectiveness of different neural networks in predicting bankruptcy filing. Two approaches for training neural networks, Back-Propagation and Optimal Estimation Theory, are considered. Within the back-propagation training method, four different models (Back-Propagation, Functional Link Back-Propagation With Sines, Pruned Back-Propagation, and Cumulative Predictive Back-Propagation) are tested. The neural networks are compared against traditional bankruptcy prediction techniques such as discriminant analysis, logit, and probit. The results show that the level of Type I and Type II errors varies greatly across techniques. The Optimal Estimation Theory neural network has the lowest level of Type I error and the highest level of Type II error while the traditional statistical techniques have the reverse relationship (i.e., high Type I error and low Type II error). The back-propagation neural networks have intermediate levels of Type I and Type II error. We demonstrate that the performance of the neural networks tested is sensitive to the choice of variables selected and that the networks cannot be relied upon to “sift through” variables and focus on the most important variables (network performance based on the combined set of Ohlson and Altman data was frequently worse than their performance with one of the subsets). It is also important to note that the results are quite sensitive to sampling error. The significant variations across replications for some of the models indicate the sensitivity of the models to variations in the data.

Fuzzy indexing and retrieval in case-based systems

Abstract: Case-based reasoning is a technique recently developed to alleviate limitations of the rule-based expert systems. Instead of relying solely on rules, a case-based system maintains old cases in a case base. When a new problem is encountered, the system retrieves similar cases from the case base and constructs a solution to the new problem based on existing solutions. A key issue in case-based reasoning is how to index and retrieve similar cases. In this paper, we present a new approach that integrates fuzzy set concepts into the case indexing and retrieval process. This approach has a few advantages over existing methods. First, it allows numerical features to be converted into fuzzy terms to simplify the matching process. Second, it allows cases in different domains to be comparable. Finally, it allows greater flexibility in the retrieval of candidate cases.

Cooperative problem solving and explanation

Abstract: Recent studies have pointed out several limitations of expert systems regarding user needs, and have introduced the concepts of cooperation and joint cognitive systems. While research on explanation generation by expert systems has been widely developed, there has been little consideration of explanation in relation to cooperative systems. Our aim is to elaborate a conceptual framework for studying explanation in cooperation. This work relies heavily on the study of human-human cooperative dialogues. We present our results according to two dimensions; namely, the relation between explanation and problem solving, and the explanation process itself. Finally, we discuss the implications of these results for the design of cooperative systems.

Using ethnography in the design of an explanation system

Abstract: In order for knowledge-based explanation systems to be acceptable, they must be useful and understandable to users. This implies that first, they should satisfy users' information needs and take account of their perspective; and second, they should be able to engage in dialogue with users. Much more progress has been made toward meeting the second condition than the first. Systems are still being produced that can engage in dialogue with users, but whose design reflects no systematic investigation of what people actually want (or need) to know about a given domain. Such lack of attention to the information needs of potential users is bound to limit the utility of any system. Because this issue relates to the concerns of social science as well as artificial intelligence, social scientists can help designers address it. One way of obtaining reliable data on the needs and characteristics of future users is ethnography, an anthropological method for gathering data in complex real-world settings. This article discusses some of the ways in which ethnography can contribute to the design process, drawing examples form an ongoing project to build an explanation system in migraine. Four aspects of our experience in using ethnography in the design process are discussed: rethinking basic design assumptions, investigating information needs, addressing the problem of perspective, and developing explanatory material. Based on this experience, the article concludes with the suggestion that the concept of explanation needs to be broadened still further to include more types of knowledge in the dialogue.

RULES: A simple rule extraction system

Abstract: We present R ULES, a simple inductive learning algorithm for extracting IF-THEN rules from a set of training examples. We also describe the application of RULES to a range of problems, each involving a different number of attributes, values, and classes. The results obtained demonstrate that in spite of its simplicity RULES is at least comparable to other inductive learning algorithms, if not more accurate and more general.

A Neural Network for Transportation Safety Modeling

Abstract: Accidents serve as an operational measure of marine safety, and specifically the safety of vessels, crews, and cargoes. The ability to accurately predict the type of vessel accident with such input variables as time, location, weather, river stage, and traffic could significantly reduce marine casualties by alerting port authorities and navigation groups as to the likelihood of a specific kind of casualty. In this paper, three models were developed to predict vessel accidents on the lower Mississippi River. These models are a neural network, multiple discriminant analysis and logistic regression. The predictive capability for vessel accidents of a neural network is compared with multiple discriminant analysis and logistic regression. The percent of “grouped” cases correctly classified is 80% (36 of the 45 cases in the testing set) for the neural network, if nonclassified cases are treated as incorrectly classified by neural network. The percent of “grouped” cases correctly classified by this network is 90% (36 of 40 cases) if nonclassified cases are excluded from the calculation. Discriminant analysis and logistic regression were able to correctly classify only 53% and 56% respectively, of accident cases into three casualty groups: collisions, rammings, or groundings.

Knowledge acquisition from multiple experts: Problems and issues

Abstract: Expert (or knowledge-based) systems are used today either as stand-alone or in conjunction with other computer-based information systems (CBIS) in thousands of organizations worldwide. Often these systems emulate situations in which the advice of several experts is needed or the knowledge required for the systems is so complex that different expert opinions are needed to establish and verify the knowledge base. Also, experts may be geographically dispersed so that it is difficult to find times to meet together. For all these reasons, knowledge engineers may need to acquire knowledge from multiple experts. The purpose of this article is to review the various problems and issues, including approaches to using experts, tools and techniques, and the benefits and limitations from working with multiple experts.

What do patients want to know: An empirical approach to explanation generation and validation

Abstract: This article describes an empirical, user-centred approach to explanation design. It reports three studies that investigate what patients want to know when they have been prescribed medication. The question is asked in the context of the development of a drug prescription system called OPADE. The system is aimed primarily at improving the prescribing behaviour of physicians, but will also produce written explanations for indirect users such as patients. In the first study, a large number of people were presented with a scenario about a visit to the doctor, and were asked to list the questions that they would like to ask the doctor about the prescription. On the basis of the results of the study, a categorization of question types was developed in terms of how frequently particular questions were asked. In the second and third studies a number of different explanations were generated in accordance with this categorization, and a new sample of people were presented with another scenario and were asked to rate the explanations on a number of dimensions. The results showed significant differences between the different explanations. People preferred explanations that included items corresponding to frequently asked questions in study 1. For an explanation to be considered useful, it had to include information about side effects, what the medication does, and any lifestyle changes involved. The implications of the results of the three studies are discussed in terms of the development of OPADE's explanation facility.

Task analysis interview based on task ontology

Abstract: In order to realize automated knowledge acquisition, we have to solve many problems such as lack of reusability and sharability of knowledge, which is one of the shortcomings in the current knowledge base technology, to fill the conceptual gap between the computer and domain experts and so on. Recent research activities in knowledge acquisition community are focused on task ontology, because it is expected to contribute a lot to making it easier to elicit expertise from domain experts. The authors have been involved in the research of knowledge acquisition and knowledge reuse. This article is concerned with task ontology and its use in a task analysis interview system MULTIS. We first discuss the knowledge reusability to identify appropriate task ontology. Then, we introduce a two-level mediating representation that contributes to bridging the gap and hence to making the task analysis interview fluent. MULTIS has been implemented in Macintosh Common Lisp.

Knowledge elicitation and knowledge representation in a large domain with multiple experts

Abstract: This paper describes the knowledge elicitation and knowledge representation aspects of a system being developed to help with the design and maintenance of relational data bases. The size algorithmic components. In addition, the domain contains multiple experts, but any given expert's knowledge of this large domain is only partial. The paper discusses the methods and techniques used for knowledge elicitation, which was based on a “broad and shallow” approach at first, moving to a “narrow and deep” one later, and describes the models used for knowledge representation, which were based on a layered “generic and variants” approach.

The value of explanation in expert systems for auditing: An experimental investigation

Abstract: Providing explanations for recommended actions is one of the most important capabilities of expert systems (ESs). The nature of the auditing domain suggests that ESs designed for audit applications should provide an explanation facility. There is little empirical evidence, however, that explanation facilities are, in fact, useful. This paper investigates the impact of explanations on changes in user beliefs toward ES-generated conclusions. Grounded on a theoretical model of argument, the study utilized a simulated expert system to provide three alternative types of ES explanations: trace; justification; and strategy. Ten expert and ten novice auditors performing an analytical review task evaluated the outputs of the system in a laboratory setting. The results indicate that explanation facilities can make a system's advice more agreeable and hence acceptable to auditors, and that justification is the most effective type of ES explanation to bring about changes in auditor attitudes toward the system. In addition, the results suggest that auditors at different levels of expertise may value each explanation type differently.

Understanding expert systems success and failure

Abstract: Many expert systems implementations are unsuccessful: the system falls into disuse or is not used to the extent originally envisioned. This paper reports on a study on the factors that lead to successful (or conversely less successful) expert system implementation. Six cases, drawn from three high-technology companies, were investigated: three very successful, three less so. Following analysis of the cases, propositions were developed regarding various success factors and their interrelation. We found that top management support and immediate manager acceptance are important, and that demonstrable business benefits and problem urgency affect management support. At the user level, perception of management support, degree of organizational change, organizational support, and users' personal stake in the system affect operational use. We relate our findings to what is known about implementation in Management Science and Information Systems, concluding that successful expert systems implementation exhibits properties found in both.

The impact of knowledge and technology complexity on information systems development

Abstract: This paper reports the findings of afield study on managing the development of software applications used for decision making. The study is based on a sample of 108 systems that have been operational for at least one year. Collectively, these systems represent a broad spectrum of complexity with respect to decision making and computer technology. At one extreme are stand-alone systems with simple decision-making logic. At the other extreme are systems with logic for highly complex decision domains. Some systems are widely distributed throughout firms and linked to suppliers, distributors, or customers. The study gathers data regarding the origins of systems ideas, development costs, project durations, management controls, and the composition of the software development teams. It develops measures to assess and categorize systems in terms of two dimensions of complexity: decision making and computer technology. Successful approaches to systems development are found to be contingent on these two dimensions of complexity.

ES∗: An expert systems development planner using a constraint and rule-based approach

Abstract: Numerous expert systems (ES) applications have been developed in the field, yet at this time no systematic, computerized tool for ES development planning exists. In this article, we present an ES development planner that adopts the constraint and rule satisfaction problems framework. We represent knowledge concerning ES development planning with objects, constraints, and rules under multiple objectives. We also present a unified reasoning process to obtain a consistent ES development plan, which has the following features: (1) supporting user interaction to reflect user intention about the problem situation and to resolve conflicts between objectives; (2) concurrent reasoning with multiple starting points to enhance the search efficiency; and (3) integrated reasoning encompassing the backward chaining popularly used in the rule-based systems and constraint propagation methods developed for constraint satisfaction problems. A prototype system ES ∗ is developed using the tool UNIK and LISP language on a SUN 4 SPARC workstation, and an application on tax advisory expert system development planning is illustrated.

Supporting learning through active retrieval of video stories

Abstract: This paper describes how a computer program can support learning by retrieving and presenting relevant stories drawn from a video case base. Although this is an information retrieval problem, it is not a problem that fits comfortably within the classical IR model of Salton and McGill. In the classical model the computer system is passive: it is assumed that the user will take the initiative to formulate retrieval requests. A teaching system, however, must be able to initiate retrieval and formulate retrieval requests automatically. We describe a system, called SPIEL, that performs this type of retrieval, and discuss theoretical challenges addressed in implementing such a system. These challenges include the development of a representation language for indexing the system's video library and the development of set of retrieval strategies and a knowledge base that together allow the system to locate educationally relevant stories.

Using rule induction for knowledge acquisition: An expert systems approach to evaluating material errors and irregularities

Abstract: There has been a significant increase in the magnitude of material errors discovered in financial statements during the 1980s. Auditors, financial analysts, and regulators have shown considerable interest in evaluating and predicting these material errors. This paper describes the development and validation of a prototype expert system, ERRORXPERT, which evaluates material errors and potential fraud. This prototype system is designed to assist auditors at the planning stage in the design of subsequent substantive tests, when material errors and irregularities in the financial statements are probable. A commercial machine learning program was used for rule induction. A set of training examples comprising error and non-error firms was used to generate rules and a separate holdout sample was used to validate the expert system results. The performance of the expert system was also compared to that of a multiple discriminant analysis model using the same data. The results demonstrate that the expert system, ERRORXPERT, outperforms the discriminant model and is a powerful evaluation tool to classify firms into error and non-error categories. The size of the sample used in this study somewhat limits the generalizability of the specific rules.

Using pattern analysis methods to supplement attention directing analytical procedures

Abstract: How might the application of analytical procedures be improved given the inherent shortcomings of traditional analytic techniques and the apparent difficulties auditors have in combining all critical cues when evaluating the results of the analytical procedures? This research attempts to improve analytical methods by applying a new technology, Artificial Neural Networks (ANNs), to perform pattern recognition of the investigation signals generated by analytical procedures. ANNs, a type of artificial intelligence technology, are able to recognize patterns in data even when the data is noisy, ambiguous, distorted or variable. Four years of audited financial data from a medium-sized distributor were used to calculate five commonly applied financial ratios. The performance of these ratios, applied independently and in combinations, was evaluated using a presumed lack of actual errors and certain seeded material errors. The ANN method evaluated the information content of the combinations of financial ratios using an entropy cost function derived from information theory. This exploratory study suggests that the use of an ANN to analyze patterns of related fluctuations across numerous financial ratios provides a more reliable indication of the presence of material errors than either traditional analytic procedures or pattern analysis, as well as providing insight to the plausible causes of the error. Preliminary results suggest that the use of pattern analysis methods as a supplement to traditional analytical procedures will offer improved performance in recognizing material misstatements within the financial accounts.

A framework for expert system implementation

Abstract: An important key to the acceptance and use of expert systems in organizations lies in effective implementation. This paper presents a framework that can be used to assess the relative importance of different factors to expert system implementation success. Since not all expert systems are alike, a typology is first presented that can be used to categorize expert systems into one of four types according to primary attributes. The literature is then used to identify a set of key factors and activities previously found to influence implementation success. Each factor is then discussed in relation to the four types of expert systems. Finally, this discussion is used to develop a framework identifying the relative importance of different implementation factors for each expert system category.

Knowledge acquisition tool for case-based reasoning s ystems

Abstract: Knowledge acquisition is an important aspect of intelligent systems. The usefulness of such systems depends on system completeness and consistency. Case-based reasoning is a useful means of matching a natural human mode of dealing with complex problems through systematic recording of experience. However, this approach requires that representative case sets be considered in the knowledge acquisition phase, and once these case sets have been acquired, an efficient means of retrieving them is needed. M-CLASS is an ordinal classification model that gives structure to a knowledge acquisition environment where cases could be the source of knowledge. The M-CLASS system assures consistency and provides a means of checking for completeness. The system is demonstrated with a medical example.

Generating explanations in context: The system perspective

Abstract: Explanations for expert systems are best provided in context, and, recently, many systems have used some notion of context in different ways in their explanation module. For example, some explanation systems take into account a user model. Others generate an explanation depending on the preceding and current discourse. In this article, we bring together these different notions of context as elements of a global picture that might be taken into account by an explanation module, depending on the needs of the application and the user. We characterize each of these elements, describe the constraints they place on communication, and present examples to illustrate the points being made. We discuss the implications of these different aspects of context on the design of explanation facilities. Finally, we describe and illustrate with examples, an implemented intention-based planning framework for explanation that can take into account the different aspects of context discussed above.

Building a fuzzy expert system for electric load forecasting using a hybrid neural network

Abstract: This paper presents the development of a hybrid neural network to model a fuzzy expert system for time series forecasting of electric load. The hybrid neural network is trained to develop fuzzy logic rules and find optimal input/output membership values of load and weather parameters. A hybrid learning algorithm consisting of unsupervised and supervised learning phases is used for training the fuzzified neural network. In the supervised learning phase, both back propagation and linear Kalman filter algorithms are used for the adjustment of weights and membership functions. Extensive tests have been performed on a 2-year utility data for the generation of peak and average load profiles in 24 h, 48 h, and 168 h ahead time frame during summer and winter seasons. From the simulation results, it is observed that the fuzzy expert system using the Kalman filter-based algorithm gives faster convergence and more accurate prediction of a load time series.

Using anthropological interview strategies to enhance knowledge acquisition

Abstract: Knowledge engineering is the “bottleneck” in expert systems (ES) development that has prevented widespread ES application. The bottleneck problem is exacerbated by a lack of specific information in published accounting research and accounting information texts describing specific knowledge elicitation activities. That lack of information adversely impacts the ability to develop cost effective ES. A review from the social sciences indicates that knowledge elicitation has been well developed by anthropologists and ethnologists. Application of anthropological interview strategies can overcome this “bottleneck.” Ford and Wood (1992) developed a strategy fof knowledge elicitation based on their research into anthropology and ethnology. Ford and Wood's (1992) four phase strategy addresses the issues of knowledge organization, problem representation, problem solving strategies, and tacit knowledge to improve the elicitation process. This four phase interviewing strategy was applied to an accounting and finance environment and a prototype ES was developed. Examples of the questions used to perform each knowledge elicitation task are presented along with a description of the controls used to assure accuracy in developing the related knowledge base. Using this elicitation strategy, allows one to overcome many of the problems previously identified with knowledge elicitation.

Toward a framework for developing knowledge-based decision support systems for customer satisfaction assessment: An application in new product development

Abstract: Customer-oriented product development has become increasingly necessary for competitive reasons. This paper describes a framework and a methodology for the design, development, and implementation of knowledge-based decision support systems for customer satisfaction assessment. A generic approach is presented that integrates knowledge-based systems with both a well-known and accepted modeling technique (scoring models), and several decision support techniques (such as the analytic hierarchy process and discriminant analysis). In addition to the fexibility and developmental advantages of knowledge-based systems, additional benefits of this approach include reduced information processing and gathering time, improved communications with senior management, and better management of scarce development resources. To simplify the exposition, we illustrate the framework and methodology within the context of a successful system implementation. The resulting system, known as the Customer Satisfaction Assessment System (CSAS), is designed to provide the decision support necessary to evaluate whether or not full-scale development of a candidate product should proceed. The system assesses and estimates the extent to which a potential new product will meet the expectations of the customer. CSAS incorporates market research findings, as well as strategic evaluation factors and their interrelationships. It can function as a stand-alone system or in conjunction with other evaluation systems (e.g., those providingfinancial, technological, manufacturing, and marketing evaluations) to provide a complete assessment of the product under consideration. Since its implementation, the experts' and other users' expressions of complete satisfaction and commitment to the system has been an indication of its value as an important decision support tool. The paper concludes with a discussion of the lessons learned for future implementations and some important extensions of this research.

Lifenet: A knowledge-based decision support tool for the risk assessment of adolescent suicide

Abstract: This paper reports on the development of Lifenet, a knowledge-based decision support system (KBDSS) for the risk assessment of adolescent suicide. Lifenet combines aspects of expert system and decision support technology, and applies them to the social service domain. Assessment is a core task in the social service industry, and its complex nature, combined with the growing use of paraprofessionals as caseworkers, provides several likely domain areas for KBDSS technology. Lifenet's main functions are to provide paraprofessional caseworkers support in reaching unbiased and consistent risk assessments, and to recommend a course of action on the part of the caseworker. The KBDSS was tested by social service professionals using actual cases, and in a controlled experiment with paraprofessionals. Paraprofessionals using Lifenet made more accurate assessments in less time than did paraprofessionals using manual assessment tools. This paper discusses the background, design, implementation, and validation of Lifenet, and presents ideas for future work.

A bond rating expert system for industrial companies

Abstract: This article presents an expert system designed to perform the task of bond rating for industrial companies. ArBor-I (articulate bond rater for industrial companies) has been developed using a production system in the environment of Common Lisp. In this article, the knowledge engineering processes involved in the development of ArBor-I are discussed. First, the nature of bond rating is explained from the viewpoint of professional bond raters. Experiential knowledge of bond raters rather than formal financial theories is captured and represented in the system. Second, knowledge acquisition sessions are briefly discussed. Protocol analysis is primarily used as a methodology of knowledge acquisition. Also, the manuals published by the bond rating companies were analyzed with the advise of expert bond raters. Third, a task analysis is conducted by identifying and conceptualizing computational generic tasks of bond rating. The task analysis subsequently specifies the types of knowledge to be captured and represented in ArBor-I. Fourth, a bond rating model is conceptualized by identifying the generic tasks and process structure. Finally, an explanation capacity of ArBor-I is discussed with demonstration of some example sessions. Computer implementation issues are also discussed followed by current limitations of the ArBor-I system.

CLXPERT: A Rule-Based Scheduling System*

Abstract: In this paper, we discuss the development and implementation of an expert system, called CLXPERT, for class scheduling at colleges or universities. Scheduling is based on the computation of a desirability map. Conflicts are resolved by using a breadth-first search in conflict trees. Information about classes and teachers are collected as the input to the system and the system outputs a class schedule that meets the specified requirements and constraints. The system also allows changes to an existing schedule to be done interactively. Rules used in the system are extracted, and reasonably modified, from the knowledge of an expert staff. The CLXPERT has been implemented in CLIPS, and has succeeded in produ the results are discussed.

Mediating between hearer's and speaker's views in the generation of adaptive explanations

Abstract: In this article, we examine the case of a system that cooperates with a “direct” user to plan an activity that some “indirect” user, not interacting with the system, should perform. The specific application we consider is the prescription of drugs. In this case, the direct user is the prescriber and the indirect user is the person who is responsible for performing the therapy. Relevant characteristics of the two users are represented in two user models. Explanation strategies are represented in planning operators whose preconditions encode the cognitive state of the indirect user; this allows tailoring the message to the indirect user's characteristics. Expansion of optional subgoals and selection among candidate operators is made by applying decision criteria represented as metarules, that negotiate between direct and indirect users' views also taking into account the context where explanation is provided. After the message has been generated, the direct user may ask to add or remove some items, or change the message style. The system defends the indirect user's needs as far as possible by mentioning the rationale behind the generated message. If needed, the plan is repaired and the direct user model is revised accordingly, so that the system learns progressively to generate messages suited to the preferences of people with whom it interacts.

Knowledge-based systems verification: A machine learning-based approach

Abstract: This paper addresses the problem of verification of knowledge bases It presents a knowledge-based system (KBS) verification approach that considers system specifications and, consequently, knowledge bases to be partially described when development starts This partial description is not necessarily perfect, and our work aims at using machine learning techniques to progressively acquire new knowledge and then improve the quality of expert system knowledge bases (KB) by coping with two major KB anomalies: incompleteness and incorrectness The KBs considered in our approach are expressed in different formalisms

Case-based retrieval of software components*1

Abstract: A major problem concerning the reusability of software is the retrieval of software components. Different approaches have been followed to solve this problem. In this paper we present the Reuse Assistant, a hybrid approach to support the retrieval of software components from a library of object classes. The Reuse Assistant consists of two subsystems that follow two different approaches: information retrieval techniques based on statistical methods, and knowledge-based techniques using some of the representation and indexing mechanisms found in case-based systems. The Information Retrieval approach grants system extendibility, and permits the use of a natural language interface. The Case-Based approach enables reasoning about concepts, allowing the retrieval of “approximate” components. Both subsystems can be operated from a common interface, where free-text and form filling queries can be posed.