Research On Service Recommendation And Reputation Management In Mobile Electronic Commerce

In recent years, with the rapid development of wireless communication technology, mobile electronic commerce plays a more important role especially for the B3G/4G communication systems.As a new business model, m-commerce has become an important component of the national economy and social information. However, immature personalization recommender system and security threats bring a negative influence on m-commerce.Therefore,to promote the development of m-commerce, it has an important theoretical and practical significance to discuss recommender system and security approach.Traditional single-criterion ratings have been successful in a number of personalization recommender systems.The single criterion, however, cannot satisfy consumers'demands. Using multi-criteria ratings instead, this paper describes a novel recommender system relying on the multi-agent technique. In the system, ratings on a service with respect to the three criteria are aggregated into an overall ranking list using a rank aggregation algorithm.Furthermore, to process single-criteria ratings, the association relationships between services are captured by using the ant colony algorithm.Based on a scenario, we describe a prototype of recommender system.Successful applications have demonstrated the performance of the system.In addition, the security requirements of services are different in m-commerce.But strong authentication for all services brings a negative influence on the quality of service since excessive computations.The balance of quality and security is an important issue.Hence, this paper presents an identity authentication using biometric identification technologies and reputation management system. Reputation management system is utilized to decide the authentication strength. Furthermore,to prevent the complexity of biometric authentication affecting efficiency,the combination of biometric authentication and PIN authentication is adopted. The combination is based on user's reputation. This novel reputation management system is able to evaluate and update the reputations of users according to the transaction history. Finally, the scenario is validated by results of various experiments that show the proposed system can be highly effective in m-commerce.