Towards the Parallel Execution of Rules in Production System Programs

Ishida, Toru; Stolfo, Salvatore

To improve the performance of forward chaining production systems, a new parallel execution model is proposed, which fires multiple rules simultaneously on multiple processor systems Two problems are discussed on the model, and efficient algorithms to solve these problems are proposed The synchronization analysis algorithm determines the necessity of synchronization between rule firings, and the decomposition algorithm determines a mapping of rules on multiple processor systems. Evaluation results on an existing production system show that a speed-up of 75 is obtained by introducing the parallel firing mechanism.



More About This Work

Academic Units
Computer Science
Department of Computer Science, Columbia University
Columbia University Computer Science Technical Reports, CUCS-154-85
Published Here
October 31, 2011