An Agent-based Improved Task Decomposition Model for Multiple Expert Systems

Abstract

To solve task decomposition of multi-expert system, agent-based task decomposition model and algorithm based on dependency is introduced. Firstly, AND/OR dependency graph for operation set is presented. Formal description of task decomposition based on With AND/OR dependency graph, is proposed. Secondly, task decomposition algorithm based on operation dependency is analyzed to solve communication among agents. Analysis of algorithm performance and instance are discussed, collaboration work flow between expert agents are analyzed and  can solve the collaboration problem among expert-agents very well. For illustration, expert system for oil-gas reservoir protection is utilized to verify the effectiveness of the method. Finally, multi-expert system based on our proposed algorithm shows that our proposed model has good time and space efficiency, and support task dynamic decomposition and can effectively solve the collaboration of experts in the multiple-expert system and improve the accuracy of inference.

DOI: http://dx.doi.org/10.11591/telkomnika.v11i10.2539