Design and implementation of distributed expert systems: On a control strategy to manage the execution flow of rule activation

摘要

There is a rich body of work dedicated to expert systems. However, none of them focus on control strategies suitable for distributed environments. We describe a novel approach to design distributed expert systems that is able to control rule activation; it involves control strategies supporting selective inferencing with rules that propagate concurrently and incorporates a set of meta-rules that operate on a blackboard and that are expressed as an assurance case. It is implemented on top of ERESYE, a tool for the realization of expert systems that is written in Erlang. We describe an implementation of the approach in an industrial setting, using an example focusing on the feature identification problem, whose resolution is an important task for source code maintenance and evolution.