An efficient approach for rule matching in production system based on multi-agent
Jiaqi Yao · 2016
Production system, which is also named as rule engine, has been widely used in artificial intelligence. Rete algorithm is the most popular rule-match algorithm, which speeds up the whole procedure by sharing the conditions among the productions and storing the temporary results. However, when the size of rules and facts continues to grow, Rete algorithm becomes computationally expensive and slow. In this paper, we propose an efficient approach for rule matching in production system based on Multi-Agent, which treats each node in rete algorithm as an agent, and every agent individually performs its own computational tasks and shares the state by passing messages. We also introduce indexes on tokens to accelerate the match and design a mechanism to balance the computational tasks among agents. Finally we conduct experiment to show its efficiency.