Explicit control of logic programs through rule algebra
Tomasz Imieliński, Shamim A. Naqvi · 1988
In this paper we argue with a basic premise in logic programming research that the meaning of a program can be inferred from its syntax alone. We show that users may have a variety of intended models for programs and that a single program may give different intended models under different assumptions of semantics. Our conclusion is that it is impossible to infer the intended model from the syntax of the program and no single semantics will capture all the intended models. We propose as a solution an explicit specification of control. Towards this purpose we define a rule algebra. The user formulates a program as an algebraic specification that directs the execution towards the intended model. The interesting question at that point is how to efficiently implement such programs. We show a natural and easy transformation such that it takes as input an algebraic specification and produces as output a program belonging to a subclass of locally stratified programs. Moreover, there is a homomorphic correspondence between the algebraic expressions and their translations.