The Web Personalized Recommender System Based On Voting Mechanism

With the popularization of the Internet and the development of digital information on enterprises, the application of E-commece is becoming more and more abroad and important. An E-commecial web site is not only a portal opening to out for an enterprise, but also a market place. E-commecial web sites are providing more and more choices for user, but their structures are becoming more and more complex. This made it hard for users to seek products or services successfully. Recommender systems can interact with users directly, provide products recommendations simulating salers, help users find useful products and finish purchasing process. And personalized recommender technologies are used more and more abroad.At present, there are more and more personalized recommender systems(PRSs) based on voting mechanism, such as sale web sites on films, books, musics and so on. In these systems, registered user obtains personalized recommendations through votings on a series of items. How to rapidly develop more clearly, more reusable and more maintainable web applications is the goal of enterprise-applications development. J2EE related technologies are preferred to realize this goal. Among the solution projects on visualization of results, light wight BIRT technology is maturing. So, the implementation of web PRSs based on voting mechanism and technologies, is becoming a hot topic for researchers.The paper introduces the concepts and classification of recommender systems, the concepts of web PRSs based on voting mechanism and common recommender technologies. It gives an emphasis on main realization technologies of web PRSs and chart analysis based on BIRT technology. And through the author's design and analysis on web PRSs, the paper used advanced development frameworks and technologies to realize a prototype of web PRS - votePRS. The paper realized visualization decription based on BIRT technology, two recommender algorithms based data mining algorithms to generate recommender results, and finished evaluation of recommender system. Through recommender analysis on standard datasets, votePRS has high performence and good results.