APL2 and AI: a study of search

Abstract

The APL2 language and its environment offers a number of features which make it an effective vehicle for AI research and for the development of knowledge-based systems. In the language, general arrays can easily model various knowledge representations; it is easy to model parallel behavior; there are various ways to implement deductive reasoning: and there are some interesting features that support a functional style of programming. In the environment, editors, interactive graphics, and debugging tools provide excellent human interfaces; Fortran, REXX, and Assembler routines may transparently replace APL2 functions; relational tables which are subsets of APL2 arrays may be retrieved dynamically; and user-to-user shared variable processing can be used to model parallel behavior. In this paper, we will demonstrate the usage of APL2 for AI problems by presenting concise functional programs for searching a database of rules and facts. Various simple enhancements are shown which extend the scope of the programs to and/or, heuristic, bi-directional, and parallel search.