ADDKnowledge-A Tool of semiautomatic Acquisition of Knowledge for the Construction of Specialist Systems

摘要

Knowledge-based systems have successfully been used to assist users to accomplish tasks in various domains from credit evaluation to aircraft design. The strength of these systems lays on the quality of the acquired and modeled domain knowledge. Knowledge acquisition has been the bottleneck of knowledge based system's development. Many studies have focused on methodological, ontological and automation issues related to knowledge acquisition. However, they present a myopic view of the problem looking at the knowledge base creation separately from the system's interface conception. This paper presents a knowledge acquisition model, ADDKnowledge, that integrates knowledge base and system's interface constructions. Our model focused on engineering design task using a parametric dependency network to represent domain (both engineering and interface domains) knowledge and a rational decision making procedure to represent the inference mechanism. In addition to presenting a model, a prototype system illustrates the feasibility of the model. The ADDKNOWLEDGE project is in its early stages. We have spent a great deal of effort building a tool to assist users to develop their knowledge base integrating three major skills: textual explanation, case examples and case experimentation. Our current effort is on maturing our knowledge acquisition model and on including interface construction assistance.