Foundations for a method for knowledge analysis

This thesis develops and justifies a method for acquiring expert knowledge and representing this knowledge as sets of practical arguments. The methodology involves an iterative analysis, wherein prior interview analysis cycles provide data for and guide subsequent interview analysis cycles. Argument structures derived from self report protocols are analyzed and hidden premises of those arguments are derived from those argument analyses. Subsequent analyses are structured by treating the hidden premises of the practical arguments as hypotheses to be confirmed. Over the course of repeated cycles the knowledge base and rule structure expands thereby increasing the constraint on the interpretation of data from subsequent interview analysis cycles.

Data was collected in two interviews with a clinical psychologist. The two interviews were subjected to a complete analysis. The second interview was structured by hypotheses formulated on the basis of data derived from the first interview. A selection of argument structures, complete with hidden premises, were then represented in the formalism of the predicate calculus.

Upon analysis of the two interview transcripts and the rule structures generated by the application of the methodology to them it has been shown that the method successfully locates hidden premises in such a way that these premises can be used to generate, in a subsequent interview, a more detailed and constrained representation of the rule structures of an expert.