Coupling committed and trial binding resolution engines

Abstract

Sequential and parallel Prolog's ability to implement don't know non-determinism in resolution efficiently makes them apt for supporting a knowledge base querying capability. On the other hand, their search based execution model their trial use of bindings makes them unsuitable for systems programming. Conversely, the use of concurrent logic programming or CLP languages of committed bindings, general and-parallelism, and synchronisation constraints enables them to support systems programming applications quite well. However, their use of committed choice resolution makes them unable to support a knowledge base querying capability. These limitations of the main logic programming schemes make it awkward to program concurrent knowledge based systems in logic. That would require the means for sustaining both knowledge base querying and systems management capabilities within a single logic programming scheme. However, the querying parts could be realised with trial binding computations on a multi-threaded Prolog engine and the systems programming parts with committed binding computations on a CLP engine by loosely coupling these engines together. This paper proposes suitable interfaces and methods for realising these forms of coupling. A multi-user Prolog system, a coupled meta-interpreter and multiple solutions predicates illustrate coupled resolution engine programming.