An agent system for the gathering of product fault information

This paper describes an agent-based system for automating the collection and processing of product fault related knowledge. The agents consist of multiple modules that individually perform simple tasks but collectively exhibit complex behavior. The agents themselves are able to interact with one another in order to process the information. Knowledge gathered by the agent system is employed to augment an existing knowledge base that in conjunction with an expert system is used for fault identification. This is achieved by means of an incremental learning algorithm. The agent system is an improvement on the conventional labor intensive and potentially error-prone procedures for the gathering and integration of new knowledge for fault identification systems