Research in the area of combining knowledge-based systems (KBSs) and database management systems (DBMSs) has traditionally followed one of three strategies. Some investigators have enhanced one class of systems with the complementary attributes of the other. Supplementing systems with the capabilities of different, existing applications, however, entails a great deal of redundant programming. Other researchers have focused on developing a new, integrated paradigm with the functionalities of both KBSs and DBMSs. This approach is revolutionary and involves a long-term basic-research effort. We have adopted a third strategy, one of coupling free-standing knowledge-based and database systems. This approach offers a practical compromise. It allows us to exploit the respective strengths of existing applications, while creating decision-support systems with improved capabilities. In particular, our use of the relational model for the DBMS component provides maximum flexibility of access by KBSs. We use a semantic data model on top of the relational model to confer application-independent semantics on the normalized data. The semantic data model facilitates the structuring and abstraction of DBMS entities into view objects, the constructs of an object-based coupling layer that we have defined. View objects subsequently can be presented to a KBS in a semantically meaningful format. In this manner, we create a KBS-DBMS environment that allows sharing of persistent data among multiple applications, while supporting the application-dependent structuring and abstraction needed by knowledge-based systems.