Top-Down beats Bottom-Up for Constraint Extensions of Datalog

This paper proposes an efficient method for evaluating queries over constraint databases. The method is based on a combination of top-down resolution with memoing and closed form bottom-up evaluation. In this way top-down evaluation terminates for all queries for which the bottom-up evaluation also terminates. The main advantage of the proposed method is the direct use of partially instantiated queries without the need for rewriting of the original program. The evaluation algorithm automatically propagates the necessary constraints during the computation. In addition, top-down evaluation potentially allows the use of compilation techniques developed for compilers of logic programming languages, which can make query evaluation very efficient.