How Does an Expert System Get its Data? (Extended Abstract)

An Expert System (Es) is a problem-8olving computer that system incorporates enough knowledge in Borne speciali2ed problem domain to reach a level of performance comparable to that of a human expert. In the heart of an expert system lie8 the program that "reasons" and make8 deduction8 ("inference engine"). To reason, knowledge both of general rules (e.g. if a person work8 for a company then he/ehe zts employee benefits) and of swzic declarative fact8 (e.g. jOhn work8 for nyu) is needed. With few exceptione, little attention is given in Es8 to the handling of very large populations Of 8PeCifiC facts. Since early prototype BSs represented specific fact8 which were characterized by large variety and a very Sma11 population, the inefficiency of data handling was not an issue. As ES8 increase in sophistication and ambition, they deal with application8 requiring a very large population of facts, often in the form of existing databases manipulated by generalized DBMS. This short paper (see [Vassiliou et al 19831 for more details) investigates the technical issue8 of enhancing expert systems with database management facilities in four stages, leading to the coupling of the ES with a large DBMS. Our vehicles are first-order logic (with Prolog) and relational databaae management.