Defeasible dependencies for relational database design

The merger of artificial intelligence techniques with database management concepts has sparked much interest. One important area related to relational databases is handling uncertain and imprecise information. However, the majority of schemes center around query processing applied to relations containing null attribute values or unknown attribute values in the tuples. That is, approaches dealing with data after the database has been designed and implemented. A much smaller amount of work has been done which addresses the design process. Typically, traditional design is based solely on absolute knowledge. The approach presented in the paper brings a new dimension to the traditional theory of relational database design. Specifically, it investigates the foundations of a theory of defeasible dependencies.