Newton C. Ellis

Knowledge conceptualization tool

Knowledge acquisition is one of the most important and problematic aspects of developing knowledge-based systems. Many automated tools have been introduced in the past, however, manual techniques are still heavily used. Interviewing is one of the most commonly used manual techniques for a knowledge acquisition process, and few automated support tools exist to help knowledge engineers enhance their performance. The paper presents a knowledge conceptualization tool (KCT) in which the knowledge engineer can effectively retrieve, structure, and formalize knowledge components, so that the resulting knowledge base is accurate and complete. The KCT uses information retrieval technique to facilitate conceptualization, which is one of the human intensive activities of knowledge acquisition. Two information retrieval techniques employing best-match strategies are used: vector space model and probabilistic ranking principle model. A prototype of the KCT was implemented to demonstrate the concept. The results from KCT are compared with the outputs from a manual knowledge acquisition process in terms of amount of information retrieved and the process time spent. An analysis of the results shows that the process time to retrieve knowledge components (e.g., facts, rules, protocols, and uncertainty) of KCT is about half that of the manual process, and the number of knowledge components retrieved from knowledge acquisition activities is four times more than that retrieved through a manual process.

Manager Associate (IKBS)

The knowledge-based Manager Associate, which assists managers in planning, organizing, staffing, scheduling, measuring, visualizing, and controlling software development processes, is described. The Manager Associate operates in a distributed Unix environment in which the manager has access to the software developer workstations through a computer network. The Manager Associate helps managers anticipate problems, which allow early corrective action. As a result, software projects can be developed on time, within budget, and to customer satisfaction. >

Expert system building tools

An overview is presented of features in several expert system building tools, and the mechanical properties of knowledge representation in these tools are described. A knowledge canonical form based on knowledge representation schemes found in commercial expert system building tools is then described. The knowledge canonical form can be used for knowledge base translation, cooperation between expert systems, and reuse of expert systems. A number of complex expert systems built from simpler loosely coupled systems are then described.<<ETX>>

A model for reengineering legacy expert systems to object-oriented architecture

Abstract The migration of existing systems to object-oriented technology is becoming increasingly important. In this paper, a reengineering model is presented. The goal of the model is to provide a comprehensive method to reengineer non object-oriented systems into object-oriented architecture. The model consists of three main processes: Reverse engineering, merging, and object-oriented development. Reverse engineering extracts requirements and knowledge from an existing software system and redocuments the system. In the merging process, recovered requirements and knowledge from the reverse engineering process are merged with new requirements and knowledge. The merging process removes redundancy, checks for inconsistency, and detects incompleteness. In the object-oriented development, a reengineered system is developed using an object-oriented software development method. This research demonstrates that successful reengineering to object-oriented architecture can be achieved by using requirements and knowledge from the original system as a basis for developing the object-oriented system. The model proved to be useful where a paradigm shift is needed. The effectiveness of the model was demonstrated by converting a legacy non object-oriented software system (implemented in C) into an object-oriented system (implemented in Smalltalk). A set of tools was built to support the model. We also identify guidelines that facilitate the transformation of legacy software systems into object-oriented systems.

The management of uncertainty in commercial expert system building tools

The uncertainty management schemes (UMSs) found in commercial expert system building tools (ESBTs) are analyzed. Important issues characterizing UMSs are considered, and UMS approaches in general are surveyed. Specific criteria which categorize the UMSs of ESBTs are presented, and uncertainty schemes in several popular ESBTs are described. The acquisition of certainty values is also discussed. Possible enhancements to UMSs in ESBTs are suggested.<<ETX>>

The visual detectability analysis associate

Abstract The outcome of this work is a knowledge-based associate ergonomic tool that assists in determining visibility of stationary small objects critical to the work environment. The Luchlesh and Moss detection solid, a design handbook guideline, accounts for non-search visibility in terms of three fundamental variables — visual angle, brightness and contrast. To reduce the difficulty of using the solid in its current form, this effort first quantified the empirical relationships of the solid. The intelligent system building tool, Exsys, was then employed: (1) to convert the results into rules for a knowledge base, and (2) to generate a tool called the Visual Detectability Analysis Associate. An efficient, convenient and easy to use ergonomic design tool is now available that determines visibility of small stationary objects in work environments where viewing time may be limited, but visual search is not required. In current use by a company to solve visibility problems encountered in system design, the VDAA tool has demonstrated beneficial savings in both time and effort.

A Comparative Study of Seven Segment Numerics

Two studies are reported. In the first study, reading seven segment numerics (SN) was shown to be more difficult than reading of conventional numerics (CN) under short viewing times. Errors in reading SN were idiosyncratic to certain numbers; however, the numbers erroneously reported are predictable. In the second study, training was shown to be effective in overcoming difficulties in reading SN, but without opportunity for continued exposure to SN, acquired skills may be reduced significantly after one month. It was concluded that where time is not critical, SN readability is comparable to CN, but in situations where time is critical, periodic SN training should be employed to preclude readability problems.

Heat Strain Assessment for Workers Using an Encapsulating Garment and a Self-Contained Breathing Apparatus

Abstract This study evaluated the physiological response of subjects wearing a Tyvek[rgrave]-Saranex[rgrave] totally encapsulating suit and a self-contained breathing apparatus (SCBA) while working under four environmental conditions. The work was performed on an ergometer connected to a measuring unit that maintained the work load at 255 W (220 kcal/h). Five young male cyclists (mean age = 28.7 years), familiar with the use of a SCBA and an encapsulating suit, were scheduled to perform this work under all four environmental conditions in a randomized series. A control session without the SCBA and the suit was performed at the end of the experiment in a cool environment. Each session was conducted in an environmental chamber and was scheduled to last 90 minutes, that is, three periods composed of 25 minutes of work each, followed by a 5-minute rest period. Physiological measurements included heart rate, six-point skin temperature (arm, chest, back, neck, lateral thigh, and medial thigh), and rectal temper...

Development of intelligent design associates: a case study

A number of human factors data sets provides guidelines and recommendations for the system design process. Much of this information is available to the human factors engineer in design handbooks, textbooks, and periodicals. While human factors design information is both available and abundant, it is not necessarily readily accessible or in a format that is easily assimilated into a particular design scenario. This paper discusses the feasibility of incorporating human factors design data into intelligent, knowledge-based systems referred to as design associates. Successful implementation of any design associate is dependent upon two key issues: conversion of human factors data into a valid knowledge base, and overcoming the knowledge representation constraints imposed by an intelligent software tool. Detailed results of a recent effort to address these issues, and subsequently implement a design associate on a commercially available intelligent software tool, are presented. This particular associate, implemented in a rule-based tool environment, aids in the selection of visual display technologies.

Knowledge sharing between distributed knowledge based systems

A distributed knowledge-based system (KBS) designed to help computer network users and administrators use, manage, and maintain computer networks is described. The initial version of the system emphasizes the types of local area networks that are typically found in large industrial, governmental, or educational organizations. Knowledge used by the system is elicited from experts, generated from logical and physical network connectivity topology, and created by analyzing data gathered by simple network management protocol (SNMP) agents, system log agents, error message daemons, and interexpert system communication daemons. Knowledge is then shared by transferring chunks of knowledge between loosely coupled KBSs. Examples of knowledge sharing between KBSs are given.<<ETX>>