Qualitative reasoning: modeling and simulation with incomplete knowledge
Qualitative reasoning: modeling and simulation with incomplete knowledge
| Hi-index | 0.00 |
The domain expert's heuristics in the performance design and improvement of collaborative systems are systematized as Qualitative rules. A knowledge base system built by using the Qualitative rules diagnoses the bottlenecks in the system operation and suggests improvements or tuning. However, the bottlenecks cannot be completely resolved due to the constraints imposed on the system resources. In such a case, our strategy is to resolve the local bottlenecks, while maintaining the global stability of the system. To enforce this strategy, we demarcate the system model into local, upstream and downstream sections, and set up the Qualitative improvement rules for each of these sections. A performance improvement task is then guided by a set of local, forward propagation, backward propagation and global rules. Local rules correspond to a local improvement. When the local rules fail to yield satisfactory results, the backward propagation rules are called upon to indicate the improvements to be made in the upstream of the network so as to resolve the problematic local bottleneck. The forward propagation rules estimate the effects on the downstream of the network due to the resolution of the chosen bottleneck. Finally, the advisability of resolving a local bottleneck, while keeping the global adverse effects to a minimum, is given by the global rules.