R1-Soar is an attempt to overcome the limitations of both expert systems and general problem solvers by combining knowledge-intensive programming based on R1 with the general weak-method problem-solving architecture of Soar. R1 is a large rule-based expert system that configures Digital’s VAX-11 and PDP-11 computer systems. Soar is a problem-solving system in which all problem-solving activity is formulated as an attempt to satisfy goals via heuristic search in problem spaces. A problem space consists of a set of states and a set of operators that transform one state into another. Starting from an initial state, the problem solver applies a sequence of operators in an attempt to reach a state that satisfies the goal.

The investigators wanted to see if a general problem-solving architecture can work at the expert system end of the problem-solving spectrum, if such a system can integrate basic reasoning and expertise, and if such a system can acquire knowledge. Using a portion of the R1 expert system with the Soar problem-solving architecture, they provide evidence that all three of these hypotheses are true.

The paper first provides an overview of R1 and how it handles the computer configuration task, and an overview of Soar. Then the structure of the combined R1-Soar is described. The final section provides experimental results and discusses the results. The paper is clearly written and has adequate references.