A natural language front-end for knowledge acquisition

Abstract

Knowledge Acquisition has historically been a bottleneck in the development of knowledge based systems. While efficient and powerful techniques have been developed for knowledge representation and processing, transferring knowledge from the problem domain into the knowledge base is characterized by the inordinate effort required by knowledge engineers; a lack of any substantial automation of the knowledge acquisition process; and in particular, the difficulty and undue complexity of the knowledge acquisition interfaces for many knowledge based systems. This paper proposes a Natural Language Interface or Front-End as a means of effective knowledge acquisition for knowledge based systems. This would have the advantage of transferring the burden of knowledge acquisition from the user to the system, while providing ease of natural language interaction to the user. By constraining such a front-end to use within limited problem domains, many of the difficulties associated with natural language processing may be avoided, while reaping the benefits of more effective knowledge acquisition. It is also recognized that the knowledge acquisition activity is one that possesses expertise of itself, and using a knowledge acquisition expert system in parallel to this front-end will likely offer further advantages in the development of part or wholly-automated knowledge acquisition.