TRICS: transformation system of relations into conceptual structure

We present an overview of a schema transformation system TRICS, and describe its procedures for schemer transformation. This system captures a conceptual structure of an existing relational database by analyzing data dependencies of relations. To capture the conceptual structure, the system extracts conceptual objects from the relations by decomposing them so that one relation represents an inherent conceptual object. The system relates the conceptual objects with three kinds of links: is-a, is-part-of, and is-associated-with; these links represent generalization, aggregation, and association, respectively. Finally, the system transforms the relational schemas of relations into frames, and constructs a frame-based knowledge base which describes the conceptual structure of the database. Frames are organized into a generalization hierarchy based on is-a links, and some frames are connected by is-part-of and is-associated-with links. Such a structural representation of an existing relational database is useful for redesigning the database, and for integrating the database with a frame-based knowledge based system.