Towards deductive object-oriented databases based on conceptual graphs

Abstract

Deductive object-oriented databases (DOODBs) are an integration of deductive databases (DDBs) and object-oriented databases (OODBs). DOODBs could be considered to be a database system which is based on logic and object-oriented paradigm. Application areas of DOODBs include advanced information systems, natural language processing and knowledge bases. Conceptual graphs (CGs), a system of ordered-sort logic, have useful constructs which are suitable for the requirements of DOODBs. This paper employs conceptual graphs to develop a foundation for DOODBs. The DOODBs are characterized by data abstraction through objects, object identifiers, object types, type hierarchy, property inheritance, methods and message passing and a logical formalism with a sound inference system. Some restrictions and extensions are proposed for the general CGs so that they can be used to represent the DOODB concepts. These extended CGs are called deductive object-oriented conceptual graphs (DOOCGs). The object types, individual objects and object identifiers of DOODBs map into concept types, individual CGs and individual referents, respectively. Methods are defined using conceptual schema graphs with bound actors and interpreted in a success/failure paradigm. A set of extended derived rules of inference is formulated for DOOCGs which are proved to be sound. The semantics of DOOCGs is also briefly outlined. A DOODB is then defined to be a set of DOOCGs together with a set of axioms and a set of inference rules.