Knowledge acquisition research supports the generation of knowledge-based systems through the development of principles, techniques, methodologies and tools. What differentiates knowledge-based system development from conventional system development is the emphasis on in-depth understanding and formalization of the relations between the conceptual structures underlying expert performance and the computational structures capable of emulating that performance. Personal construct psychology is a theory of individual and group psychological and social processes that has been used extensively in knowledge acquisition research to model the cognitive processes of human experts. The psychology takes a constructivist position appropriate to the modelling of human knowledge processes, but develops this through the characterization of human conceptual structures in axiomatic terms that translate directly to computational form. In particular, there is a close correspondence between the intensional logics of knowledge, belief and action developed in personal construct psychology, and the intensional logics for formal knowledge representation developed in artificial intelligence research as term subsumption, or KL-ONE-like, systems. This paper gives an overview of personal construct psychology and its expression as an intensional logic describing the cognitive processes of anticipatory agents, and uses this to survey knowledge acquisition tools deriving from personal construct psychology.

Knowledge acquisition tools based on personal construct psychology

In many practical knowledge engineering contexts, interview data is the commonest form in which information is obtained from domain experts. Having obtained interview data the knowledge engineer is then faced with the difficult task of analysing what is initially relatively unstructured and complex material. It is argued that the knowledge engineer's task of analysing interview data conceptually as part of the knowledge elicitation process is similar to that of the social scientist analysing qualitative data. One implication of this is that a range of methods originally developed by social scientists for the analysis of unstructured and semi-structured qualitative material will be of assistance to the knowledge engineer. The background philosophical issues linking qualitative social science research and knowledge elicitation are outlined; both are characterized as fundamentally creative, interpretative processes. “Grounded Theory”, a social science methodology for the systematic generation of conceptual models from qualitative data, is described in detail. An example is presented of the use of Grounded Theory for the analysis of expert interview transcripts, drawn from a knowledge engineering project in civil engineering. The discussion focuses upon the processes used to move from an initial unstructured interview transcript to a core set of interrelated concepts, memos and models that fully describe the data.

The use of grounded theory for conceptual analysis in knowledge elicitation

Knowledge engineering for industrial expert systems

The knowledge engineer is only weakly supported at three critical stages in the knowledge engineering life cycle: (1) knowledge acquisition during which problem conceptualization must largely be tackled with paper and pencil; (2) knowledge encoding, during which it is frequently necessary to be able to navigate across a variety of knowledge representation formalisms; and (3) large-scale debugging, in which the graphical rule traces cannot cope with enormous rule sets involving hundreds or thousands of rules. The research described attempts to provide just such support through complementary visual programming (VP) and program visualization (PV) techniques embedded in a fully implemented software environment called KEATS: the knowledge engineer's assistant. Several novel visual programming and program visualization techniques aimed at knowledge engineers have been developed, which include (1) a hypertext transcript analyzer from which conceptual models can be generated, (2) a direct graph manipulation sketchpad which allows the knowledge engineer to sketch out objects and relations (including control flow and rule dependencies) from which code can be generated, and (3) dependency viewers which allow the knowledge engineer to examine and manipulate temporal and logical rule dependencies at different levels of granularity. How these facilities are incorporated into KEATS and the key themes that emerge from this approach to visual knowledge engineering are discussed. >

Visual knowledge engineering

Knowledge acquisition as a process of model refinement

The ‘Knowledge Engineer's Assistant’ (KEATS) is a software environment suitable for constructing knowledge-based systems. In this paper, we discuss its role in supporting the knowledge engineer in the tasks of knowledge elicitation and domain understanding. KEATS is based upon our own investigations of the behaviour and needs of knowledge engineers and provides two enhancements to other modern ‘shells’. ‘toolkits’, and ‘environments’ for knowledge engineering: (i) transcript analysis facilities, and (ii) a sketchpad on which the KE may draw a freehand representation of the domain, from which code is automatically generated. KEATS uses a hybrid representation formalism that includes a frame-based language and a rule interpreter. We describe the novel components of KEATS in detail, and present an example of how KEATS was used to build an electronic fault diagnosis system.

Kent Pitman

Papers

Support for knowledge acquisition in the Knowledge Engineer's Assistant (KEATS)