Testing implications of data dependencies

ABSTRACT

We present a computation method---the chase---for testing implication of data dependencies by a set of data dependencies. The chase operates on tableaux similar to those of Aho, Sagiv, and Ullman. The chase includes previous tableau computation methods as special cases. By interpreting tableaux alternately as mappings or as templates for instances, we can test implication of functional and join dependencies. This information is useful in determining when a relational database scheme accurately represents the information it is intended to. The chase can also be used to test equivalence of database schemes and as part of the test of whether the relation schemes in a database scheme are independent components.