The relational production model of problem solving, due to Vere, is used to analyze plan inefficiency. The concept of plan variant is introduced and precisely related to the triangle tables of Nilsson. The concept is used to extend certain classes of inefficiencies. A method is described which "tightens up" plans containing these inefficiencies, often producing optimal ones.

Logic programs offer many opportunities for parallelism. We present two models of computation which allow parallel processing of Prolog programs. In the goal list model each processor is given a separate approach to the problem. In the AND/OR model each processor is given a separate subcomputation. Each model allows an arbitrary number of processors, and both perform the same sequence of unifications as would the standard depth first interpreter for Prolog if only one processor were available. In each model the parallelism is achieved as a result of the non-determinism in the program.

Logic programs offer many opportunities for parallelism. We present an abstract model that exploits the parallelism due to nondeterministic choices in a logic program. A working interpreter based on this model is described, along with variants of the basic model that are capable of exploiting other sources of parallelism. We conclude with a discussion of our plans for experimenting with the various models, plans which we hope will lead eventually to a multi-processor machine.

This paper is an attempt to explain apparent confusion of efforts in the field of artificial intelligence (AI) research in terms of differences between underlying methodological perspectives held by practicing researchers. A review of such perspectives discussed in the existing literature will be presented, followed by consideration of what a relatively specific and usable taxonomy of differing research perspectives in AI might include. An argument will be developed that researchers should make their methodological orientations explicit when communicating research results, both as an aid to comprehensibility for other practicing researchers and as a step towards providing a coherent intellectual structure which can be more easily assimilated by newcomers to the field.

A system is described which learns to compose sequences of operators into episodes for problem solving. The system incrementally learns when and why operators are applied. Episodes are segmented so that they are generalizable and reusable. The idea of augmenting the instance language with higher level concepts is introduced. The technique of perturbation is described for discovering the essential features for a rule with minimal teacher guidance. The approach is applied to the domain of solving simultaneous linear equations.

The goal of the proposed research is to develop methods for efficient implementation of logic programs. There are two areas we wish to investigate, both of which are continuations of research conducted by members of the UCI dataflow architecture group. One aspect of the proposed research involves development of a non-von Neumann architecture for parallel execution of logic programs; preliminary work in this area is reported by Conery [9]. The second area invovles transformation of high level logic specifications into efficient Prolog and/or procedural language programs, and is based on work by Morris [20].