Knowledge Engineering Techniques Research Articles

Representing Expert Knowledge for Instructional System Design: A Case Study

This paper describes the process by which an AI knowledge engineering technique was used for performing knowledge acquisition within the context of Instructional System Design. The method is a type of cognitive task analysis in which conceptual graphs are developed using several complementary knowledge acquisition methods. The graphs can be used to explicitly represent a variety of knowledge types including semantic, rule, and implicit knowledge. We outline the process by which we applied the method to represent expert knowledge for instructional design in the domain of engineering mechanics, and describe modifications and lessons learned from the application.

Progress in knowledge engineering for image interpretation and classification

Knowledge engineering techniques in the field of expert systems have common grounds with pattern recognition and can be applied for image interpretation and classification. The effects of the degeneration of the method of statistical pattern recognition into a rule base are studied in a case study on soil classification from SPOT images. Implementations of a Boolean and a Maximum Likelihood Classifier are demonstrated using an “expert tutor” based on a home developed PROLOG shell with binary logic and one using a commercial shell (ENVISAGE).

Expert Systems in Control Engineering: A Review of the Perspective

With the increasing complexities of modern day plants and systems as well as the explosion of the control engineering packages and techniques, the use of expert systems is gaining popularity. A number of expert systems has been built and applied to solve control engineering problems. This paper attempts to present a perspective view of this new area. A number of important issues such as numerical stability of control algorithms, different knowledge engineering tools and techniques and their application to different control engineering expert systems has been addressed in this paper. The salient features of a number of representative systems have also been discussed.

A TOOL FOR CONSTRUCTING INTERACTIVE SOFTWARE

The Tailor is a research tool for refining and evaluating a design representation for software systems that have a high degree of textual and graphical interaction. The design representation incorporates both software engineering and knowledge engineering techniques, providing a set of conventions for describing software design information and software construction knowledge. The Tailor demonstrates that the design representation, when embedded in a productivity tool, overcomes limitations of current generation user interface tools. For example, the Tailor has the capability to foresee construction difficulties arising from earlier design decisions. It uses this capability to provide active guidance in producing a software design that avoids problems otherwise overlooked during software specification and design.

Representation of models for expert problem solving in physics

A computer program, APEX, is proposed to investigate idealized formal models representing physics problems. Two types of models are defined: canonical physical objects and physical models. During problem solving, the problem is represented as a data connection network, which is progressively augmented by these models in the form of additional network elements. APEX employs views as a representational framework for connecting the initially informal objects to the formal models of the domain. The view framework supports multiple representations (e.g., viewing many objects as a single canonical physical object), handling of incompletely specified problems, and invertibility of the views. This computational framework provides a powerful representational mechanism that allows a finite set of physical principles to be applied to a potentially infinite variety of problems. As a knowledge engineering technique, views allow general principles to be applied to a variety of objects whose representations differ. >

An improved design procedure for fuzzy control systems

The purpose of this paper is to introduce simpler and more effective method in designing fuzzy controllers. A scale for partitioning the universe of discourse and choosing the appropriate fuzzy set shapes for the control variables is introduced and is termed “fuzzimetric arcs”. Knowledge engineering techniques are used to obtain process information which is interpreted using systematic analysis as an aid to the design of fuzzy logic controller. This has been illustrated by a manufacturing system example in the form of a welding application which provides a guide to the reader who is unfamiliar with such techniques.

原子炉異常診断エキスパートシステム「ＤＩＳＫＥＴ」の異なる記述言語による性能比較

An expert system DISKET has been developed at JAERI to apply knowledge engineering techniques to the transient diagnosis of nuclear power plant. The first version of DISKET programmed in UTILISP has been developed with the main-frame computer FACOM M-780 at JAERI. The LISP language is not suitable for on-line diagnostic systems because it is highly dependent on computer to be used and requires a large computer memory. The large mainframe computer is also not suitable because there are various restrictions as a multi-user computer system. The second version of DISKET for a practical use has been developed in FORTRAN to realize on-1ine real time diagnoses with limited computer resources.These two versions of DISKET with the same knowledge base have been compared in running capability, and it has been found that the LISP version of DISKET needs more than two times of memory and CPU time of FORTRAN version. From this result, it is shown that this approach is a practical one to develop expert systems for on-line real time diagnosis of transients with limited computer resources.

Expertise

Expert systems in nursing are developed with traditional knowledge engineering techniques. These techniques focus on the behavior and logic of the expert, not qualities of expertise. Expertise has been described but not explained. This article proposes a theoretical framework for the study of expertise that can be used to facilitate the development of expert systems.

An Automated Control System-Designing System Assisted by Hypothetical Reasoning

Most studies of the so-called CACSD (Computer-Aided Control System Design) have aimed at the development of computer systems which will assist control engineers to design effectively, making better use of sophisticated machine interfaces and the graphic environment, and do not aim at automated control design beyond assisting the user in designing control systems. The present study has developed a system, which is aimed at the automation of classical control system design, by describing the dialogue between man and computer as a computer program through techniques of knowledge engineering. In particular, hypothetical reasoning systems supported by TMS (Truth Maintenance System) are used in order to automate trial and error procedures which are problems in classical design methods. In other words, a hierarchical system has been constructed where a control system-designing expert system based on hypothetical reasoning is placed at the meta-rule layer above the procedure and tool layers. This paper illustrates the basic structure of this expert system and the utilization forms of hypothetical reasoning in control system design, and shows the actual operation of the system as seen through a design example.

Experimental Evaluation of a Diagnostic Rule-Based Expert System for the Nuclear Industry

This paper summarizes an experiment that was conducted at the international research institute, the OECD Halden Reactor Project, Halden, Norway to assess the effectiveness of the diagnosis expert system DISKET (Diagnosis System Using Knowledge Engineering Technique) for helping nuclear reactor operators diagnose and treat malfunctions. The objectives of the experiment are described, the experimental design and the methods of data analysis are presented, and the results of the analysis are discussed. In summary, the DISKET system improved the performance of test subjects in performing the actions needed to treat the malfunction, but the effects on the diagnostic processes were not conclusive. The improvement in treatment performance is encouraging in itself, but the question of how this occurs has not been completely answered, because of the difficulty in measuring human performance for cognitive activites such as diagnosis. The need for improved methods for measuring diagnostic performance in future experiments is discussed.

A knowledge-based architecture for protein sequence analysis and structure prediction

Methods for analyzing the amino-acid sequence of a protein for the purposes of predicting its three-dimensional structure were systematically analyzed using knowledge engineering techniques. The resulting entities (data) and relations (processing methods and constraints) have been represented within a generalized dependency network consisting of 29 nodes and over 100 links. It is argued that such a representation meets the requirements of knowledge-based systems in molecular biology. This network is used as the architecture for a prototype knowledge-based system that simulates logically the processes used in protein structure prediction. Although developed specifically for applications in protein structure prediction, the network architecture provides a strategy for tackling the general problem of orchestrating and integrating the diverse sources of knowledge that are characteristic of many areas of science.

A large project for demonstrating knowledge engineering techniques including applications of neural networks

A large project for demonstrating knowledge engineering techniques including applications of neural networks

A knowledge-based environment for modeling and simulating software engineering processes

The design and representation schemes used in constructing a prototype computational environment for modeling and simulating multiagent software engineering processes are described. This environment is called the articulator. An overview of the articulator's architecture identifying five principal components is provided. Three of the components, the knowledge metamodel, the software process behavior simulator, and a knowledge base querying mechanism, are detailed and examples are included. The conclusion reiterates what is unique to this approach in applying knowledge engineering techniques to the problems of understanding the statics and dynamics of complex software engineering processes.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

The guide to available mathematical software advisory system

The primary goal of the Guide to Available Mathematical Software (GAMS) project is to provide convenient access to information about mathematical and statistical software which is available to computer users at the National Institute for Standards and Technology. The principal vehicle through which this information is disseminated is an on-line advisory system called the GAMS Interactive Consultant. In this paper we describe the current status of the GAMS project. We then enumerate some of the weaknesses of the system and suggest knowledge engineering techniques which may alleviate them.

Artificial intelligence knowledge acquisition techniques for instructional development

Knowledge engineering techniques for developing expert systems may also be useful for instructional development. A review of knowledge engineering focusing on knowledge representation and knowledge acquisition suggests ways in which these methods could be adapted to developing instructional systems. As further work is done on intelligent computer-assisted instructional systems and other complex instructional development projects, knowledge engineering skills may become more important for the instructional developer.

Plant Operator Performance Evaluation System

A plant operator performance evaluation system to analyze plant operation records during accident training and to identify and classify operator errors has been developed for the purpose of supporting realization of a training and education system for plant operators. A knowledge engineering technique was applied to evaluation of operator behavior by both event-based and symptom-based procedures, in various situations including event transition due to multiple failures or operational errors The system classifies the identified errors as to their single and double types based on Swain's error classification and the error levels reflecting Rasmussen's cognitive level, and it also evaluates the effect of errors on plant state and then classifies error influence, using “knowledge for phenomena and operations”, as represented by frames. It has additional functions for analysis of error statistics and knowledge acquisition support of “knowledge for operations”. The system was applied to a training analysis for a scram event in a BWR plant, and its error analysis function was confirmed to be effective by operational experts.

Heuristic framework for the resource constrained multi-project scheduling problem

Real world scheduling problems can be affected by diverse and conflicting goals. Some scheduling problems are bounded with limited resources and represent a considerable challenge to planners. When multiple projects are involved, scheduling problems become even more complex and difficult to resolve. Because of the combinatory explosions and unrealistic assumptions, traditional management sciences techniques such as PERT, CPM, and a host of similar project schedulers are very limited to special cases of scheduling problems. Recent developments of artificial intelligence (AI) and knowledge engineering techniques have made the development of expert systems which can be driven by scheduling heuristics to resolve problems associated with the traditional optimization techniques by giving “better” rather than “best” solutions. This paper presents the conceptual framework and development strategy for an expert system in multi-project scheduling domains. This paper will also present a practical application of the scheduling system.

Plant operator performance evaluation system.

A plant operator performance evaluation system to analyze plant operation records during accident training and to identify and classify operator errors has been developed for the purpose of supporting realization of a training and education system for plant operators. A knowledge engineering technique was applied to evaluation of operator behavior by both event-based and symptom-based procedures, in various situations including event transition due to multiple failures or operational errors The system classifies the identified errors as to their single and double types based on Swain's error classification and the error levels reflecting Rasmussen's cognitive level, and it also evaluates the effect of errors on plant state and then classifies error influence, using “knowledge for phenomena and operations”, as represented by frames. It has additional functions for analysis of error statistics and knowledge acquisition support of “knowledge for operations”. The system was applied to a training analysis for ...

The development of power system restoration method for a bulk power system by applying knowledge engineering techniques

A prototype expert system for bulk power system restoration is discussed. Restoration of a bulk power system involves operation of equipment that is complexly inter-related. For this reason, object-oriented programming techniques using a frame representation of knowledge and heuristics has been adopted to achieve efficient processing. The use of knowledge representation and processing is described, and the restoration process is examined. A prototype system that has successfully satisfied the requirements of speed, stability and flexibility is described. >

The Development of Power System Restoration Method for a Bulk Power System by Applying Knowledge Engineering Techniques

The Development of Power System Restoration Method for a Bulk Power System by Applying Knowledge Engineering Techniques