V. Tomasevic (Serbia and Montenegro)
Expert systems, explanation, dynamic tracking process,meta-knowledge, control mechanisms
To understand the expert system behaviour, its explanation facility provides to the user an access to as much of the system knowledge as possible. In most of the explanation systems, the tracking mechanism maintains a list of all successfully fired rules or satisfied goals. This imposes a problem of how to record all relevant data for each applied rule when too many rules are invoked. Moreover, the system has to consume additional time and space overhead for unnecessary data recording. We propose a way to improve the tracking mechanism by introducing some kind of restrictive strategy, as a means of dynamic controlling which rules or goals should be tracked during the problem solving process. Our approach is based on the use of meta-rules for managing the key parameters that define the current state of the tracking mechanism. The implementation of this idea in the EXBEST explanation system is presented.
Important Links:
Go Back
