| Abstract | The idea of a Deductive Object-Oriented database(DOODB) is to combine the concepts of deductive databases(DDBs) and object-oriented databases(OODBs) in a single database system in order to offer the advantages offered by each of them. This kind of database is suitable for knowledge representation and many advanced database applications like CAD/CAM databases, VLSI design databases, office automation, engineering design and CASE databases, where conventional databases are not suitable. A groundwork for deductive objectoriented databases based on conceptual graph has been carried out. The formalism of conceptual graphs(CGs), a knowledge representation scheme in Al, offers some interresting constructs that are particularly suitable for the requirements of merging DDB and OODB concepts. The DOODBs are characterised by data abstraction through objects, object identities, object types, type hierarchy, property inheritance, methods and message passing, a logical formalism with a sound (deductive) inference mechanism as well as a single language with declarative expressions for databases, queries, integrity constraints and answer to queries. Some restrictions and extensions are suggested for general conceptual graphs to form Deductive Object-Oriented Conceptual Graphs(DOOCGs), which have been used to represent the DOODBs. The object types in DOODB are represented by concept types of CG, individual objects are mapped into individual graphs of respective types, object identity is modeled by individual referents. Methods and message passing are modeled using conceptual schema graphs, actors and dataflow graphs. Methods are interpreted in a success/failure paradigm. A fixpoint semantics of deductive object-oriented conceptual graphs has been defined. A direct derivation proof procedure has been formulated for DOOCGs and it has been proved as sound and complete. A notion of Deductive Object-Oriented CG theory(DOOCG theory) has been proposed, and the DOODBs are represented as DOOCG theories. |