Focus: This paper adopts a pragmatic and retrospective stance, reporting on what has been achieved in the past decade in knowledge acquisition research based on personal construct psychology.
Introduction: Knowledge-based systems development is targeted on the emulation of human high-level skilled performance in a computer-based program (i.e., an expert system). Distinguishing features of the knowledge-based systems approach: first, it emphasizes the use of human experts as the primary sources of information; and, second, it emphasizes the use of knowledge representation schema as the primary basis of implementation. Knowledge-based system development emphasizes in-depth understanding and formalization of the relations between the conceptual structures underlying expert performance and the computational structures capable of emulating that performance. Figure 1 represents graphically the psychological and computational foundations for theories, methodologies and tools supporting expertise transfer. The most important aspect of supporting expertise transfer effectively is providing feedback to the human expert for purposes of validating that the knowledge structures at each stage of the transfer properly represent the basis of the skilled performance, and for verifying them operationally as far a posible at every stage of developmment. Much of the functionality of the tools involved in expertise transfer is targeted on supporting human understanding of the development process. Additionally, the tools supporting expertise transfer are developing and refining a model of the basis of the expertise as a specification for a system to emulate it.
George Kelly's 'repertory grid' methodology for eliciting conceptual structures has become a widely used and accepted technique for knowledge elicitation, and has been implemented as a major component of many computer-based knowledge acquisition systems. The grid itself dates back to Kelly's (1955) application of the personal construct psychology that he had developed for clinical and teaching practice. The repertory grid was an instrument designed by Kelly to bypass cognitive defenses and give access to a person's underlying construction system by asking the person to compare and contrast relevant examples. It is often easier and more accurate for the expert to provide critical cases rather than a domain ontology. Gaines & Shaw (1980) suggested that repertory grids would provide a useful development technique for expert systems, and later published a validation study of the elicitation of expertise from accountants and accounting students using computer-based repertory grid elicitation (Shaw & Gaines, 1983). Shaw (1980) took advantage of the processing power and interactivity of computers to introduce on-line analysis and feedback to the person from whom the grid was being elicited. Shaw & Gaines (1986) introduced new forms of analysis of the repertory grid based on fuzzy sets theory which became the basis of rule extraction.
Repertory Grids: Primarily a technique for building the conceptual structure of the expert without direct elicitation of concepts and their structures and relationships. The repertory grid is just a first step: collecting extensional data. The extensional specification of how concepts apply to individuals is inadequate to fully specify the concept structure. Further analysis techniques are needed to approximate the structure from the extensional data.
Using Repertory Grids: I am still a bit unclear how to select experts. For example, do I select people to whom I attribute some sort of expertise? Or, do I choose people who have demonstrated expertise and are generally accepted as experts by many? The domain I am interested in is TEACHING USING TECHNOLOGY, which is a combination of both teaching expertise and expertise using technology, however, it seems that defining this type of expert is more complex considering the levels, or gradations, of expertise that may exist. An individual who is an expert teacher, but a novice, though enthusiastic, user of technology in the classroom might not fit the category of expert teacher using technology. However, this individual displays expertise in teaching, and one might expect that characteristics of this expertise overlap, or might be predicates to, their desire to incorporate information technology in their classroom repertoire. The reverse situation also presents itself; an expert user of technology may demonstrate intermediate teaching skills. It seems that the selection of experts must be done with careful attention to the exact type of expertise one is attempting to model.
An additional question I have concerns the elicitation process itself. It seems there are several ways one could approach the elicitation task of distinguishing elements to derive constructs:
KSS0:As I read this part of the article, I wondered how extensively the students in this class will be participating in the knowledge acquisition process. Namely, which parts of the tools will we be using? Surely, we will be using ELICIT, with either/both Focus & PrinCom, as well as SOCIO, to compare the structures generated by different experts in the same sub-domain. Is one of our goals to use INDUCT, to translate the structures into concept subsumptions or rules for an expert system shell? Is a further goal to use EXPORT, to transform the knowledge analysis into formalisms understandable by knowledge-based system shells? Or, is our goal situated more in the overall development methodology, illustrated in Figure 33?
Conclusions: The authors of this paper, having presented a comprehensive review of knowledge acquisition research based on personal construct psychology, suggest that this research is still in its infancy, and that future research will add major new developments which build on revealled potential to date.