The effect of unary inclusion dependencies on relational database design

Abstract

Functional dependencies (FDs) and inclusion dependencies (INDs) are the most fundamental database integrity constraints, and they are used in many data models. In a previous paper, we described a synthesis algorithm for the design of a relation database from FDs. In this paper, the effect of unary inclusion dependencies (UINDs) on the relational database design is studied. Though the implication problem for a set of INDs and FDs is undecidable, if attention is restricted to unary INDs, there will be a complete axiomatization and its decision problem can be solved in polynomial time. To discover new FDs and INDs from a set of FDs and UINDs, an effective algorithm is presented to find k-cycles in the multi-graph presentation of FDs and UINDs. Finally, the synthesis algorithm is enhanced by considering interaction between FDs and unary INDs.