Automatic Acquisition Of Knowledge For Solving Analysis Tasks

Patel, Jitendra M (1990). Automatic Acquisition Of Knowledge For Solving Analysis Tasks. PhD thesis The Open University.

DOI: https://doi.org/10.21954/ou.ro.0000fc3d

Abstract

The extraction of knowledge from domain experts for the purpose of building expert systems has been found to be a difficult and a time- consuming enterprise. Researchers have therefore looked at the - possibility of automating this task of knowledge acquisition. The existing knowledge acquisition systems rely on either some task-specific knowledge or general techniques for knowledge elicitation for obtaining problem-solving expertise from domain experts. Generally, systems which employ task-specific knowledge produce better problem-solving expertise than those systems which use elicitation techniques. However, the scope of applicability of former systems is relatively narrow. This is the central problem addressed in this thesis: to design a knowledge acquisition system which can be applied over a wide range of tasks with the purpose of acquiring useful problem-solving knowledge.
The thesis presents a methodology and a system for knowledge acquisition called ASKE. The methodology prescribes that knowledge acquisition should start by defining the task and then use the developed task-model to acquire domain specific knowledge. ASKE is able to support this process by allowing the user to construct task-models and by being able to effectively use them for acquiring domain expertise. The advantage of progressing in this manner is two-fold: firstly, it widens the scope of applicability of the knowledge acquisition system; and, secondly, it makes possible the construction of knowledge-bases that exhibit expert performance.
ASKE contains knowledge engineering expertise which it uses to help domain experts encode their problem-solving expertise directly into a knowledge-base. The system derives its power from the templates which encode knowledge. The templates serve a triple function: they represent knowledge - their normal function; they encode expectations of the kind of knowledge that is to be acquired; and, they serve as a guide for how problem-solving knowledge may be organized so as to facilitate its encapsulation into a knowledge-base.