Over the years, there has been a confluence of concepts, tools and tec
hniques from two diverse areas: artificial intelligence (AI) and datab
ase (DB) systems. There are several ways to integrate expert systems (
ES) and database systems. This paper surveys the related literature an
d classifies the integrated systems into four classes: enhanced DB, en
hanced ES, coupling of existing ES and DB, and expert database system.
A new loose coupling approach (Simple Coupler) based on predefined SQ
Ls is then proposed. Its system architecture and system operations are
described. The Simple Coupler is compared with the DIFEAD (Dictionary
Interface for Expert Systems and Databases) approach and the commerci
al ES shell approach on the criteria of the independence of the DB, ES
, the complexities and future expansibility. This approach has great p
ractical values because of its simplicity. Finally, this paper reports
on two prototypes coupling existing systems: accounting DB & financia
l ES and medical history database & medical diagnosis system, respecti
vely, in the PC window environment. (C) 1997 Elsevier Science Ltd.