Eliminating combinatorics from production match

Eliminating combinatorics from the match in production systems (or rule-based systems) is important for various reasons like real-time performance, the utility issue in machine learning, parallel implementations and modeling human cognition. The goal of this thesis is to eliminate match combinatorics without sacrificing production system functionality. The thesis focuses on restricting expressiveness for achieving this goal. It explores a variety of approaches for restricting expressiveness in production systems, which engage in different trade-offs for eliminating match combinatorics. It uses Soar, an integrated problem-solving and learning system, as a vehicle in this investigation. This investigation rules out several of the approaches, but it reveals two promising approaches that could form the basis of future production systems.