Much effort has been expended on developing special architectures dedicated to the efficient execution of production systems. While data-flow principles of execution offer the promise of high programmability for numerical computations, we demonstrate here that the data driven principles can also be applied to symbolic computations. In particular, we consider a mapping of the RETE match algorithm along the line of production systems. Bottlenecks of the RETE match algorithm in multiprocessor environment are identified, based on which the algorithm is parallelized. The modifications to the actor set as well as the program graph design are shown for execution on the Tagged-Token Data-flow Computer. The results of a deterministic simulation of this multiprocessor architecture demonstrate that artificial intelligence production systems can be efficiently mapped on data-driven architectures.
All Science Journal Classification (ASJC) codes
- Data-flow principles of execution
- expert systems
- parallel processing
- pattern matching
- production systems