An Application of Fuzzy Cluster Analysis to Matchmaking in Middle Agents

Abstract

For multi-agent systems used in open environments, agents might appear and disappear unpredictably. Middle agents are advantageous since they allow a multi-agent system in open environments to operate robustly in the face of agent appearance and disappearance. The performance of middle agents relies heavily on the matchmaking algorithms used. Matchmaking is the process of finding an appropriate provider for a requester through a middle agent. It was argued that the practical performance (track record) of service provider agents has a significant impact on the matchmaking outcomes of middle agents. However, the track records of agents are accumulated gradually during the executing process of a multi-agent system. Thus there are no track records available when the system is first launched. To this end, this paper proposes an initial value generation approach, which is based on fuzzy clustering algorithms. The initial values generated can be used by matchmaking algorithms at the very beginning of a system running. Some simulation experiments were conducted. The experimental results show that the proposed approach can generate reasonable initial values for track records. With the agents' story and the initial values of the track records, the performance of matchmaking algorithms can be improved significantly.