Book Description
In the emerging Electronic Marketplace, intelligent agents that are reactive, proactive and social will become the major players. These intelligent agents will search for good deals, evaluate trading partners and eventually provide advice on or even make decisions on transactions for their principals. To this end, it is vitally important for the agents to be able to precisely estimate the trust of the business partners to make successful deals. This dissertation proposes an adaptive reputation-based trust model which can be utilized by intelligent agents in an open Electronic Marketplace, where the seller agents and buyer agents can enter and leave the market freely and the service quality and price of goods vary. The trust model is based on an agent's reputation history, witness testimony, and other weighting factors. Learning is integrated to make the trust model adaptive and robust in a dynamic environment. To verify the proposed adaptive reputation-based trust model and the significance of the constructs and factors of the model and to compare the performance of the proposed model with other models, a multi-agent system is built to simulate the interactions among these agents.
