A Uniform Framework for Deductive Database Derivation Strategies

A uniform framework is presented to describe the most typical strategies that are used to compute answers to Deductive Databases. The framework is based on the definition of a general Least Fixpoint operator that operates on meta rules. Each set of meta rules represents a different strategy, and this allows an easy comparison. We first consider Deductive Databases with Horn clauses and we present strategies based on the general ideas of Forward chaining, Backward chaining, and their combinations. We show that well known strategies, like Magic Sets, can be represented in this general framework, and can be compared to other ones. Then we consider non Horn clauses, and we present strategies to compute Intensional answers and Conditional answers. The purpose of the paper is mainly didactic even if new strategies are presented in the last section. The presented framework should make easier comparisons between existing strategies.