Knowledge-contribution motivated adaptive neighborhood for multi-agent communities
Wan Huang, David W. Cordes · 2004
In this paper we present a mechanism that facilitates adaptive neighborhood construction for multi-agent communities in the context of electronic trading. The proposed method is based on the posteriori evaluation of knowledge contribution that occurs during trust evaluation. The concept of "credit", which is derived by benchmarking a witness' opinion with the evaluating agent's own experiences and the actual benefits from the transaction, is used to represent the knowledge contribution by a witness. In our process, a hill-climbing strategy enables the agent to adaptively find neighbors who contribute the most to the trust evaluation process.