Join dependencies in relational-database design

Abstract

Problems of logical scheme design for relational database systems are considered when the integrity constraints specifying an application area are join dependencies (JD). The admissibility of an arbitrary database scheme is defined in terms of the existence of some JD. The problem of generating various admissible schemes is reduced to the inference problem for JD. A system of inference rules is presented to deduce JD logically implied by a given set of JD. To estimate the influence of a scheme on query evaluation, the problem of finding a minimum subset of database relations ensuring the correct answer is investigated. A polynomial algorithm is developed for finding a minimum set of database relations whose join provides the same answer to a given query as the join of the whole database (assuming all database relations to be projections of a single relation).