Knowledge Acquisition and Constructivist Epistemology
|1.||An automated approach may be moreefficient than manual interviewing methods, thereby reducing the great expense presently incurred in the knowledge acquisition phase.|
|2.||Automated approaches may prove able to elicit expertise not easily obtained by manual interviewing methods, thus producing systems with greater expertise.|
To provide these desired benefits, automated approaches to knowledge acquisition must assist the knowledge engineer in avoiding the domain experts' cognitive defenses and reduce the representation mismatch -- the difference between the manner in which the domain expert normally states knowledge and the way it is represented in the expert system knowledge base. The design and construction of knowledge acquisition tools have recently become areas of intense research and development.
As noted by Bradshaw and Boose ( 1990), a major difficulty has been that much of the aforementioned work lacks a plausible theoretical foundation:
As a consequence of incomplete theory and a limited repertoire of practical approaches to the dynamics of the modeling process, knowledge engineers have had to rely on intuition and experience as the primary means of developing and testing effective procedures. (p. 129)
Many of those engaged in knowledge acquisition (as researcher or practitioner) may be classified as toolmakers and/or tool users. Toolmakers should exploit theory as a means of building their tools on a sound footing and as a framework in which to make explicit their epistemological assumptions. Furthermore, theory may offer toolmakers a useful infra-