AAMAS '04 Proceedings of the Third International Joint Conference on Autonomous Agents and Multiagent Systems - Volume 2
Trust-Based Community Formation in Peer-to-Peer File Sharing Networks
WI '04 Proceedings of the 2004 IEEE/WIC/ACM International Conference on Web Intelligence
An electronic marketplace based on reputation and learning
Journal of Theoretical and Applied Electronic Commerce Research
Trust modeling for message relay control and local action decision making in VANETs
Security and Communication Networks
Hi-index | 0.00 |
In this thesis, we propose a market model and learning algorithms for buying and selling agents in electronic marketplaces. We take into account the fact that multiple selling agents may offer the same good with different qualities, and that selling agents may alter the quality of their goods. We also consider the possible existence of dishonest selling agents in the market. In our approach, buying agents learn to maximize their expected value of goods using reinforcement learning. In addition, they model and exploit the reputation of selling agents to avoid interaction with the disreputable ones, and therefore to reduce the risk of purchasing low value goods. Our selling agents learn to maximize their expected profits by using reinforcement learning to adjust product prices, and also by altering product quality to provide more customized value to their goods. We experimentally evaluate our model on both microscopic and macroscopic levels. On the micro level, we examine the individual benefit of agents, in particular their level of satisfaction. Our experimental results confirm that in both modest and large-sized marketplaces, buying and selling agents following our proposed algorithms achieve better satisfaction than buying and selling agents who only use reinforcement learning. On the macro level, we study how a marketplace populated with our buying and selling agents would behave as a whole. Our results show that such a marketplace can reach an equilibrium state where the agent population remains stable and that this equilibrium is beneficial for the participant agents. The market model and learning algorithms presented in this thesis can therefore be used in designing desirable market environments and effective economically-motivated agents for e-commerce applications.