A novel multi-agent knowledge reasoning method for cooperation and confrontation

The objective of knowledge reasoning (KR) aims to extract the effective information among the complex environment and massive data. It has been recently used in many fields such as the military operation, the network confrontation, the athletic game and so on. The problems of the low reliability, the high subjectivity, and the complex modeling always exist in many KR methods. This paper proposes a novel KR method based on the evidence theory for the multi-agent cooperation and confrontation platform. Period to utilize the proposed method, the key factors affecting the platform status are chosen as the evaluation indicators. According to the histogram method, the mass value of each indicator within every agent is calculated. For the more convenient calculation the holonomic mass value set of all agents is then divided into n sub-evidences and with the sub-evidences the information synthesis is conducted by the evidence theory for the final decision basis. The results of the simulation experiment demonstrate that the proposed method has better properties in accuracy and efficiency than the competed methods.