Knowledge-based Approach for Abstracting Relational Data into a Frame Model

In RDBMS, dynamic semantics such as categorization, aggregation and generalization cannot be recognized when mapped into a logical relational schema. Programming is needed in the database applications to preserve the semantics since the database cannot provide such enforcement. A frame model is proposed for (i) storing semantics from relational database and (ii) enforcing constraints of relational database by facilitating rule-based operations. It is a knowledge representation schema for semantics and contains constraint and method classes to enforce the dynamic semantics within the schema. It is a meta data on top of a relational database. Within the frame model, relational data semantics are captured into 4 classes. User input is needed to rebuild the domain integrity and relation integrity from the existing data. Querying data is applied to verifying user supervision by discovering hidden semantics in the databases so that the translation from relational database schemas to frame model can be done smoothly. Even though there are many possible frame models that can be constructed from a known relational database, the translated frame model is the one close to user expectation, and verified by the present database state (instances).