AbstractExpert systems are becoming increasingly complex and diverse. The need for efficient data management for their growing knowledge bases is apparent. Moreover, with the widespread use of Database Management Systems, the operational data required by the expert system is often already available from an online database. Hence the interaction of expert systems and databases is a topic which is receiving increasing attention. This paper reviews existing techniques for such interaction and describes a novel dictionary–driven approach to the problem, called a Dictionary Interface for Expert Systems and Databases (Difead). A prototype of Difead is described, which couples a Medical Diagnosis System, written in Lisp, with a relational databas