A blackboard architecture for control
Artificial Intelligence
Embedable Problem-Solving Architectures: A Study of Integrating OPS5 with UMass GBB
IEEE Transactions on Knowledge and Data Engineering
| Hi-index | 0.00 |
A distributed blackboard implementation that addresses the following network configuration problem is described. Telecommunication network managers at the Dutch PTT are regularly asked to install new connections between a group of users and a specific application. Several pieces of equipment have to be assembled. Choosing the specific pieces and types of equipment depends on many factors, and these choices are difficult because of the large number of constraints and the different kinds of equipment available. A distributed approach was chosen because of the problem domain's inherently distributed character, and a blackboard architecture because of its suitability for distributed AI systems, planning problems, and incremental system development. The system was implemented in the distributed blackboard environment, Blondie-III, as part of a larger project researching distributed AI. The problem is described in some detail, the reasons for using a distributed blackboard system to solve it are discussed, and the implementation and its evaluation are presented. The system's most striking feature is not the distribution of the problem-solving process, but the way search is conducted on the blackboard, as well as the natural integration of several types of knowledge (including rule-based and procedural knowledge).