A deductive rule-based language for object-oriented databases is presented. A deductive rule in this language derives new patterns of associations among objects of some selected classes if these objects fall in certain 'base' on other derived patterns. The patterns of object associations derived by a rule are held in a subdatabase whose intention consists of some selected classes and their associations. In other words, the structure of a derived subdatabase is represented using the structural constructs provided by the object-oriented data model and hence can be uniformly operated on by other rules to further derive new subdatabases. Therefore, the world of subdatabases is closed under this rule-based language.>

The object orientation of a semantic association model (OSAM) is presented. It integrates the concepts and techniques of semantic modeling and those introduced by the object-oriented paradigm. Unlike conventional data models such as the relational model, the object orientation of OSAM allows the user to model an application in terms of complex objects, classes and their associations, instead of tuples (or records) and relations (or record types). The primitives (objects, class, instance and link) and the perspectives (class and object) of an OSAM database are described. Key differences between OSAM and a conventional object-oriented model are discussed. The features of object orientation and explicit definition of semantic associations among objects allow the database of an application domain to be modeled, accessed and manipulated at a higher conceptual level and thus simplify the tasks of the users in the development of their applications.>