Database reasoning: extracting knowledge from databases (abstract)

As interest in developing expert systems continues to grow, it become more important for us to increase our understanding of how to successfully build these systems. The major problem that currently constricts the development of expert systems is the formidable process of eliciting knowledge from human experts. A possible solution to alleviate this knowledge-acquisition bottlenecks is to equip historical database with some deductive capabilities that will allow us to emulate expert-system reasoning entirely in the database's terms. In this paper we describe a system that tries to extend an existing database containing component test-repair history with several features to support inferencing at the reasoning level of human trouble shooters. That is, the system features automatic knowledge-base derivation. This presentation demonistrate the structure of a system that, when queried with a test step, will provide trouble shooting advice describing the most likely cause for failure.