Towards a formal semantics and inference rules for conceptual data models

摘要

A new-from a theoretical point of view-approach to expressing data- or knowledge base semantics is described: the universe of discourse is modeled by classes (unary relations), by attributes (binary relations), and by a set of general, possibly multityped functional and existence dependencies between these classes and attributes. In contrast to single-typed approaches, such as normalization, our approach has to cope with dependencies ranging over more than one type. The notation of a path and of confluent type hierarchies are introduced to cope with multityped dependencies. An axiomatization for general functional and existence dependencies is provided. Possibilities and limitations of our approach for automating reasoning involving conceptual schemas are discussed, and its relationship to relational database theory is outlined.