Showing papers in "Expert Systems With Applications in 1993"

Knowledge based information retrieval system

Abstract: An information retrieval system with good human-interface methods to give the system ease-of-use having two distinctive features with the first being visual interface and the second being natural language interpretation. The visual interface provides for one is visual interaction for local search and natural language interpretation provides for linguistic interaction for global search. The visual interface provides versatile views onto the contents of the knowledge base that the system has, controlling mechanisms for browsing through the knowledge base, a capability of showing relevant information for the users, and a mechanism for editing a query expression that describes information to retrieve. By using the visual interface for information retrieval, the users can easily create query expressions, by consulting and reacting with the system. The natural language interpretation makes use of a conceptual network as a knowledge-base that stores important concepts and relationships among these concepts. Based on knowledge and information represented in the conceptual network, the meaning of a noun phrase or a nominal compound which is a string of adjectives and nouns with some prepositions can be inferred. The inferred interpretation of such a noun phrase is paraphrased into an expression that the information retrieval system can handle. Therefore, the user of the system can simply describe the desired information in a language to get the desired information.

Integrating target analysis and tabu search for improved scheduling systems

Abstract: This paper explores the integration of the Artificial Intelligence/Operations Research approach known as target analysis with tabu search to create a more effective system for machine scheduling. Target analysis is designed to give heuristic and optimization procedures the ability to learn what rules are best for solving particular classes of problems. The authors focus on the development of rules that depend on memory functions to incorporate diversifying elements in a tabu search method which is tailored to find optimal or near optimal solutions for a class of single machine scheduling problems with delay penalties and setup costs.

Case-based reasoning: An introduction

Abstract: Case-based reasoning is a method of solving a current problem by studying the solutions to previous, similar problems. This article presents an introduction to the case-based reasoning process, including an example of the creation and consultation use of the case base. Construction tools for case-based reasoning are identified, and key concepts in case-based reasoning are discussed.

PERFEX: An expert system for interpreting 3D myocardial perfusion

Abstract: Interpreting three-dimensional (3D) data is generally recognized as an ill-defined and information-intensive task. The task becomes increasingly difficult in the context of medical diagnostic imagery, wherein the visual information must be interpreted in conjunction with other, nonvisual information. A novel approach is presented to perform the interpretation of such multidimensional information, concentrating on a medically important application: the interpretation of 3D tomograms of myocardial perfusion distribution. The overall goal is to assist in the diagnosis of coronary artery disease. The approach employs knowledge-based methods to process and map the 3D visual information into symbolic representations, which are subsequently used to infer structure (anatomy) from function (physiology), as well as to interpret the temporal effects of perfusion redistribution, and assess the extent and severity of cardiovascular disease both quantitatively. The knowledge-based system presents the resulting diagnostic recommendations in both visual and textual forms in an interactive framework, thereby enhancing overall utility. This paper presents the methodology underlying this approach, including the implementation and testing of this system within an actual clinical environment.

Verification and validation of case-based systems

Abstract: Verification and validation of artificially intelligent systems has been the focus of substantial recent research. However, little attention has been given in the literature to verification and validation for cased-based systems. The unique structure of cased-based systems is used to raised new validation issues, develop new approaches to generating comparative solutions for validation purposes, and investigate new approaches for examining the quality of the case base. In addition, this article presents new statistical and structural approaches designed to exploit unique aspects of case-based reasoning for verification purposes.

Generalized case-based reasoning system for portfolio management

Abstract: A case-based reasoning system (CBRS) is appropriate for an experince-rich domain, while a rule-based system performs reasonably well in a knowledge-rich application environment. Performance of a CBRS suffers when past experience is not readily available. A generalized case-based reasoning system (GCBRS) is proposed to remedy this weakness by incorporating domain theories represented as generalization rules. With these rules, previous experience (stored as cases) can be generalized so that the possibility of solving a new case is higher than it would be when case-based reasoning is used alone. The architecture and the inference mechanism of a GCBRS are discussed in this article. A portfolio management support system based upon the proposed GCBRS architecture is presented to demonstrate the feasibility of using GCBRS for developing a decision support system in a knowledge-poor and experience-poor domain. This article concludes with a discussion of future research.

Data input/output control system

Abstract: A batch type data input and output control system executes a file output program which outputs data to a file stored in an external storage unit from a main storage apparatus, and a file input program for processing the data in the file. An input/output parallel management means for managing parallel processing of the file output program and file input program and a data transfer unit are provided for transferring the data within the main storage unit without transferring the data through the external storage unit. The data written by the file output program is directly transmitted to the file input program, for instance in units of a character or record under control of the input/output parallel management means. The file output program and file input program are executed in parallel, and can be registered or deleted by designation of another program or a user. When one of the file output program and the file input program ends in an abnormal state, the other program is compulsorily ended. When both programs end in a normal manner and a subsequent program is designated, the subsequent program is automatically initiated.

Case-based reasoning: Market, applications, and fit with other technologies

Abstract: Case-based reasoning (CBR), the hit of the American Association of Artificial Intelligence annual conference in 1991 and 1992 is now enjoying a surge of interest in its first year of commercial availability. Knowledge-based system designers, developers, integrators, and tool vendors are now seriously considering the role and utility of CBR in leveraging the vast experience within organizations for more effective decision making. The potential market for CBR appears enormous, particularly in more complex problem-solving domains, but the areas of most immediate interest are in applications where efficient information processing needs are urgent, such as automated help desks. Early experiments pairing CBR with rule-based systems will soon lead to hybrid combinations with other “close approximation” technologies, such as neural networks, fuzzy logic systems, genetic algorithms, and so forth. CBR appears headed for a sustaining role not only as a useful complement in knowledge-based information processing technology but also as an engine for “mainstream” information tasks of the future (e.g., intelligent text processing and retrieval, data mining, and projective reasoning). This article will discuss this emerging role for CBR and its implications from a marketing perspective.

An expert system approach to AIDS intervention and prevention

Abstract: The increased involvement of information systems in Acquired Immunodeficiency Syndrome (AIDS) intervention and prevention (AIP) leads to increased demand for expert systems. This paper presents an expert system that extracts data from the database and then screens AIDS-risky behaviors and schedules follow-up assessments by using mathematical models, as intelligent as an AIP expert would. The application results have been compared with AIP expert's judgement that presents evidence of the usefulness of the system.

SIMON: A distributed computer architecture for intelligent patient monitoring

Abstract: Intelligent real-time patient monitoring encompasses data acquisition and reduction, sensor validation, diagnosis, therapy advice, and selective display of information. This paper describes the architecture and the functionality of a prototype intelligent patient monitoring system, named SIMON, designed to meet these requirements. In SIMON, the various aspects of a monitoring task are performed by three semi-independent modules running asynchronously: the feature extraction, the patient model, and the display modules. Central to SIMON is the notion of context sensitivity which permits (a) the adaptation of the monitoring strategy in response to changes either in the patient state or in the monitoring equipment and (b) the contextual interpretation of incoming data. SIMON is currently applied to the task of monitoring newborn infants with respiratory distress syndrome (RDS) and undergoing assisted ventilation.

MICRO-BOSS: A micro-opportunistic factory scheduler

Abstract: Recent research in factory scheduling has demonstrated the benefits of building schedules by first optimizing the sequencing of bottleneck machines, namely, machines whose utilizations are expected to be particularly high. Within this approach, two scheduling perspectives are generally adopted: a resource-centered perspective is used to help maximize the utilization of bottleneck machines, and a job-centered perspective is used later to compactly complete each job schedule (i.e., to reduce work-in-process inventory within each job). Because new secondary bottlenecks may arise during the construction of the schedule, recent scheduling systems have been designed with an ability to switch back and forth between their resource-centered scheduling perspective and their job-centered scheduling perspective. This ability to revise the current scheduling strategy dynamically has been termed opportunistic scheduling . However, because these schedulers require scheduling large resource subproblems or large job subproblems before revising their scheduling strategy, we refer to them as macro-opportunistic schedulers . Instead, this paper describes MICRO-BOSS, a so-called micro-opportunistic scheduler , that can revise its scheduling strategy each time an operation is scheduled. Experimental results suggest that the extra flexibility of a micro-opportunistic approach to scheduling often translates into important reduction in schedule costs.

LBS: Bayesian learning system for rapid expert system development

Abstract: This paper presents the results of experience with a novel expert system shell called Learning Base System (LBS) The primary goal in developing LBS was to enable the rapid development of testable expert systems The strategy adopted was to use a Bayesian classifier system as the form of knowledge representation, and adapt it to allow incremental acquisition-of knowledge from both data and experts, and prediction and validation procedures The advantages and limitations of the system are described in three applications The first application is the diagnosis of diseases in crops, illustrating knowledge acquisition by an expert in a data-poor domain The second illustrates how LBS could be used in a geographic information system The third is the development and testing of models for predicting wildlife density solely from data The Bayesian classifier is shown to be a flexible formalism for implementing a wide variety of knowledge-based tasks

On the applicability of diagnostic criteria for the diagnosis of rheumatoid arthritis in an expert system

Abstract: Cadiag -2/ Rheuma is a medical expert system developed to assist in the differential diagnosis of rheumatic diseases. Based on fuzzy set theory and fuzzy logic, it supports the formalization of vague and uncertain medical information (i.e., medical entities and relationships between them) and draws justifiable conclusions from these imprecise data. Given a patient's finding patter, Cadiag -2 provides confirmed and excluded diagnoses, diagnostic hypotheses, and suggestions for further examinations. The knowledge base of Cadiag -2 has been designed to contain simple finding/disease relationships as well as diagnostic rules of high complexity to confirm or hypothesize disease. We shall present results obtained with 300 clinical cases from a hospital for rheumatic diseases. Different rules for the diagnosis of rheumatoid arthritis based upon classification criteria issued by the American Rheumatism Association were tested against each other. That diagnostic rule which had shown the best results was then further improved by a rheumatology expert, which finally yielded a sensitivity of 83.3% and a specificity of 95.3%.

User interface design to expert systems based on hierarchical spatial representations

Abstract: User interface becomes an important issue as expert systems are introduced in complex real-world applications. Graphical representations of the domain model can often be used as a base of the interface, particularly in domains with strong spatial relationships which can be presented through drawings, plans, or maps. In this paper the interface of an expert system in the area of chemical emergencies management is presented. This interface is based on the Hypermap paradigm, a hypertext-like graphical representation of the accidcent site and a geographical database. User interaction is based on a direct manipulation dialogue. As a result of this work, the usability of the expert system has been considerably increased.

Contingencies for the design of scheduling expert systems

Abstract: To design a scheduling expert system under the dynamic environment, one needs to consider not only the performances of tardiness and flow time during the generative scheduling phase, but also the robustness and scheduling time requirements during the reactive control phase. To find the best contingent system's architecture (centralized or hierarchical) in combination with the scheduling strategies on activity ordering, allocation time slot selection, and activity partitioning on multiple parallel machines, a series of experiments have been conducted. The findings can be used to guide the design of scheduling expert systems.

Knowledge-based system for rostering

Abstract: Scheduling problems are commonly found in servicing and manufacturing industries. In order to maximize the utilization of the supporting hardware and machines, machine operators usually are required to work on shift so that machines remain running continuously without stopping. The construction of the shift duty assignment, usually called roster, is a nontrivial task since many factors need to be considered, such as the maximum number of hours an operator can work continuously and the minimum rest time between two consecutive duties. Since the construction of roster requires nontrivial intellectual activities, it is difficult to computerize roster scheduling using conventional data processing techniques. On the other hand, such roster planning process has been performed by the human planner for years, and experience has been collected to perform the planning task efficiently. Knowledge-based approach takes advantage of human expertise and thus is used to automate the roster scheduling activities. We report here a case study on designing and implementing a computer rostering system for a group of airport servicing staffs using knowledge-based approach. Prototyping is used in the development process. Knowledge is represented in two different forms, namely forward chaining rules and constraints. Forward chaining and iterative improvement are used as control strategies in the system. The first prototype was completed in less than a month. During the refinement process, a tool, called SHO, was developed to simplify the modification process. The whole system was completed in approximately 6 months, of which most of the time was spent on knowledge and rules refinement. Initial implementation was developed using Common Lisp on a Sun Workstation. It is then ported to PC Scheme, a Lisp dialect, on an IBM Personal Computer. The system has been used for rostering up to 500 airport servicing staffs for a few years. As most knowledge and rules have been incorporated into the system, the system has been rewritten in C recently so that the knowledge-based system can communicate with other systems in the organization and has a better performance.

MODELER: Incorporating experiences to support model formulation—A case-based planning approach

Abstract: Model formulation is key to most mathematical modeling tactics. The enormous amount of time needed to formulate problems is well documented. In this article, we focus on the three elements of model formulation—experience, planning, and opportunism. These elements are used in the implementation of MODELER, a case-based planning system that integrates case-based reasoning methods and an opportunistic control in a blackboard data architecture. An analysis of expert verbalizations provided significant inputs to the design of MODELER.

An intelligent database system for composite material selection in structural design

Abstract: This paper describes the development of a prototype intelligent database system for composite material selection in structural design. This intelligent system integrates the expert system with the database system to provide decision-making support systems that exhibit some forms of intelligence. The overall architecture of this system is illustrated. The present capabilities of this system are discussed and demonstrated with an example problem.

Information processing models for sleep staging

Abstract: Three different methods of automated sleep staging are described and compared in the paper. The interesting aspect of the comparison is that the inputs to the three different information processing models (an expert system, a belief automaton, and a neural network) are the outputs of the same fronted processor that excerts EEG features. We found out that the level of man-machine agreement is very similar among the systems (around 85%) approaching the level of agreement required among human experts. However, the similarity of performance in such a diversified set of approaches points out the need for enhancing the accuracy of the first level of analysis (EEG feature extraction).

FMS scheduling and control: Learning to achieve multiple goals

Abstract: This paper describes FMS-GDCA, a loosely coupled system using a machine learning paradigm known as goal-directed conceptual aggregation (GDCA) and simulation to address the problem of Flexible Manufacturing System (FMS) scheduling for a given configuration and management goals. The main advantage of FMS-GDCA is that it provides a manufacturing manager with an extremely flexible and goal-seeking control mechanism that allows for a continuous improvement in decision outcomes. The manager can choose a goal or a combination of goals or a combination of goals or can prioritize the partial goals by assigning weights. Given the goals, FMS-GDCA attempts to achieve them to the best of its ability. If it cannot meet the goals due to its lack of knowledge, it will acquire the relevant knowledge from data and solve the problem. The results indicate that FMS-GDCA can consistently produce improved overall performance over the traditional scheduling techniques.

Neural networks as expert systems

Abstract: This paper shows that the high level decision-making function of expert systems, that depend upon many levels of logic, can be implemented in a neural network without the engineering of a detailed knowledge structure. Further, the neural network can interpolate and extrapolate a discrete set of associated input and output vectors, so that the output decision space is continuous. A drawback to the use of neural networks for decision making is that their training is universally problematic. We simplify the process with a new random optimization algorithm that consists of a global stage and a local stage. Unlike other methods, we also optimize the exponential rise parameters α and β in the sigmoids at the middle and output layers, which increases the speed of learning and decreases the minimum sum squated error.

Acquiring implicit knowledge in a complex domain

Abstract: This article describes GDCA-II, a machine learning-based system that acquires implicit knowledge through model abstraction in a flexible manufacturing system (FMS) domain. GDCA-II employs an integrated strategy involving conceptual clustering and case-based learning to acquire knowledge relevant for solving domain problems. Threshold values and bottleneck resource examples are used to demonstrate the necessity of acquiring implicit knowledge for improved decision making. Simulation results indicate that by acquiring implicit knowledge in problem solving considerable improvements in FMS performance can be achieved.

Problem acquisition in scheduling domains

Abstract: In knowledge acquisition, knowledge engineers typically lack the background needed to pose optimal questions, to capture the right knowledge, and to structure the knowledge acquired about a particular application area. A large amount of time and effort is consumed in forming a detailed coherent model of the application domain. Even after the model is formed, it is typically informal both in their mind and in their documentation. This makes further acquisition of problem solving knowledge and later system design difficult. This paper discusses the situation and presents a model-based approach to problem acquisition in scheduling domains. In order for this approach to work, an adequate model of scheduling problems is a prerequisite. In the paper, a domain-independent model and a problem-formulation framework are described. Guided by them, knowledge engineers will know what to ask in their interview with domain experts and how to organise formally the domain knowledge they acquire.

A comparison of two intelligent scheduling systems for flexible manufacturing systems

Abstract: Despite the existence of hardware suitable for the development of advanced automated manufacturing systems, the implementation of such systems has been hampered by the lack of appropriate software necessary for the scheduling and control of these systems. Artificial Intelligence (AI) has been suggested as a methodology suited to the development of this software. As a result, in this paper two intelligent scheduling and control systems are developed with the cooperation of an “expert” at an existing FMS in Aiken, South Carolina, USA. The literature related to scheduling FMS using AI methodologies is unclear as to whether scheduling should be done in a real-time manner similar to simple job-shop scheduling or in a predictive manner that shows detailed start times and finish times for some scheduling horizon. As such, one intelligent scheduler developed in this research utilizes a real-time scheduling methodology, while the second utilizes a predictive methodology. Both systems were developed in conjuction with a scheduling expert at the FMS. Results from a simulation model of the FMS using each of the two scheduling methodologies are compared in an effort to address the issue of which methodology is better suited for the scheduling and control of automated manufacturing systems such as FMS.

Formal validation of a knowledge-based system using a variation of the Turing test

Abstract: The proliferation of computer-based information systems makes an organization's operations exceedingly vulnerable to several types of risks, including loss of data, unauthorized use of data, errors in recording data, and loss of assets or property. Exposure to such risks is typically mitigated through the institution of internal controls. Since systems designers are often not cognizant of control design strategies, the participation of an auditor has been recommended. To avoid potential auditor independence problems caused by the same set of auditors participating both in the design and evaluation of control strategies, a knowledge-based system for control design has been developed. Ten practicing internal auditors participated in the validation of the knowledge-based system using a variation of the Turing test. This paper reports on the validation procedures: results indicate that the knowledge-based support system can perform at a level comparable to that of internal auditors for the purchasing cycle in business organizations, thereby helping alleviate the auditor independence problem.

An expert system using an exchange heuristic for the resource-constrained scheduling problem

Abstract: In this paper, we discuss the development and implementation of an expert system for the resource-constrained scheduling problem. The objective of the system is to minimize the makespan (or duration of a project). This approach can accommodate up to 400 operations (activities) and 50 different resources. The knowledge contained in the expert system is based upon an exchange heuristic algorithm that has been applied succesfully to various resource-constrained scheduling problems and has been shown to be a very general and efficient search method for large-sized problems. The exchange heuristic takes an initial schedule and improves upon it via exchanging and rearranging operations, while continually maintaining feasibility. In this system, a forward chaining strategy is used as the inference process. These production system language OPS83 with external C-language links is utilized as the system development tool. In order to evaluate the performance of the expert system, we compared it with an earlier heuristic programming version and found that superior computation performance and enhanced software maintenance were achieved. A number of large-sized problems have been tested on this system, and the results have shown that the makespans were reduced by 10–16 percent compared to initial schedules. Advanced programming techniques were utilized to reduce computer memory requirements and to increase the search speed. Therefore, the system can deal with large-sized problems on PC-compatible machines (with 640KB RAM or more), running MS-DOS, with short CPU execution times.

Episodic cases and real-time performance in a case-based planning system

Abstract: In this article, we discuss COOKIE, an integrated planning, execution, and learning system that operates in the domain of meal planning and preparation. Its episodic representation scheme is used to avoid brittleness in planning, help attain real-time performance, and simplify the case acquisition process. An episodic representation of an event is one that allows reconstruction of the event—all of the actions, when they happened, what facts were true, and so forth. In COOKIE, this is a set of propositions in a temporal logic corresponding to the initial conditions, end conditions, goals, actions, and observations taken during the preparation of a meal. This representation is usable for a range of reasoning tasks: plan generation and projection, plan recognition, explanation, and failure recovery. This representation also promotes efficient real-time performance. The details present in the episode are maintained in the generated plans, providing the necessary information to schedule actions and observations before plan execution begins. We describe how our execution monitor was implemented to take advantage of this information while performing in an efficient manner.

On the representation and management of medical records in a knowledge-based system

Abstract: The need for adapting Hospital Information Systems (HISs) to new applications (aimed at clinical data management) and to the ever growing population requires an evolution from both the representation and the usability points of view. The goal of this paper is to discuss the interdependencies of the advanced data representation models necessary to the new HISs and the interaction requirements. In particular, we present the technical taken in the MILORD (Multimedia Interaction with Large Object-oriented Radiological and clinical Databases) project to implement a departmental environment which integrates different kinds of clinical data in a user-friendly and homogeneous way. The implementation platform of the MILORD system is based on results obtained in the KIWIS project, which produced an advanced knowledge-base environment for large database systems. The interaction with a HIS is analyzed identifying the HIS-specific interaction tasks, the typical phases of the interaction with an information system, and the necessary interaction paradigms. In particular we describe the approach adopted in KIWIS for overcoming the problem of querying object-oriented databases and we discuss the new requirements that multimediality poses on the interaction paradigms.

Scheduling approaches in expert systems: A study of their applicability

Abstract: Expert scheduling systems are being used in a variety of domains, including manufacturing, personnel, aerospace, and military applications. Very little work has been done in determining when it is most appropriate to use a particular scheduling approach in an expert system. This paper addresses this area, placing special emphasis on matching scheduling techniques to requirements in the NASA (National Aeronautics and Space Administration) mission planning arena.