Semantic Update Optimization in Active Databases

In an active database, an update may be constrained by integrity constraints, and may also trigger rules that, in turn, may a ect the database state. The general problem is to e ect the update while also managing the \side-e ects" of constraint enforcement and rule execution. In this paper an update calculus is proposed by which updates, constraints and rules are speci ed and managed within the same formalism. Constraints and production rules are expressed in a constraint language based on rst-order logic. These logic expressions are used to semantically transform an original update into a sequence of updates that re ect the relevant constraints and production rules. The inference mechanism associated with processing a reformulated query ensures that: 1) the preand post-conditions of an update are satis ed, 2) update side-e ects are propagated, and 3) repairs are made to tuples exhibiting constraint violations. Thus, a user-speci ed \update" is transformed, through semantic reformulation techniques, into a sequence of updates which together ensure semantic integrity of the original update as well as its propagated side-e ects. This research was supported in part by an ARPA grant, administered by the O ce of Naval Research under grant number N00014-92-J-4038. This research presents several contributions. Integrity constraints and production rules are expressed in a declarative formalism so that they may be reasoned about. The update calculus formalism handles the semantic reformulation of an update to re ect relevant constraints and rules governing it. Finally, an algorithm is presented to handle constraint enforcement, production rule ring, and subsequent repair actions.