Abstractions in Logic Programs

Most logic programming languages have the first-order, classical theory of Horn clauses as their logical foundation. Purely proof-theoretical considerations show that Horn clauses are not rich enough to naturally provide the abstraction mechanisms that are common in most modern, general purpose programming languages. For example, Horn clauses do not incorporate the important software abstraction mechanisms of modules, data type abstractions, and higher-order programming. As a result of this lack, implementers of logic programming languages based on Horn clauses generally add several nonlogical primitives on top of Horn clauses to provide these missing abstraction mechanisms. Although the missing features are often captured in this fashion, formal semantics of the resulting languages are often lacking or are very complex. Another approach to providing these missing features is to enrich the underlying logical foundation of logic programming. This latter approach to providing logic programs with these missing abstraction mechanisms is taken in this paper. The enrichments we will consider have simple and direct operational and proof theoretical semantics.