Automating the knowledge acquisition process in expert system development

With the ever growing interest in the use of expert system technology for so phisticated problem solving, it becomes increasingly more important to expedite the traditional knowledge acquisition process from which these systems evolve. The traditional process-in which a knowledge engineer familiarizes himself with the domain, helps the domain expert formalize the required knowledge, and fi nally implements the representation of this knowledge in machine usable form-is slow, cumbersome, and time-consuming. In the research reported in this thesis, an attempt has been made to automate the knowledge acquisition process by em ploying a "learning by being told" methodology. The scope of acquired knowledge ranges from domain knowledge, including the information about objects and their relationships, to the procedural knowledge associated with the functionality of the mechanisms.

AKAS (Automatic Knowledge Acquisition System) is the constructed prototype for demonstrating proof of concept in which a domain expert directly interfaces with the knowledge acquisition system to ultimately construct the knowledge base for the particular application. In this prototype, the expert talks directly to the system using a natural language restricted only by the extent of the definitions in the analyzer dictionary containing a subset of concepts related to a given domain. In this case, the domain is the electrical system of the BOEING 737. AKAS uses a set of seed concepts and vocabulary to capture information. Using this information, the system extracts and differentiates the incoming knowledge into two fundamental classes - knowledge related to physical objects and their relationships to each other in the domain, and heuristic and functional knowledge about mechanisms. These two classes of knowledge are represented in the form of frames and functions, respectively, to create the knowledge base and to augment the control strategy of the expert system.