Study On The Intelligent Commodity Recommended System Based On Products Gene And GA

With the rapid development of electronic commerce, website emerged into a problem of merchandise information overload, which is very difficult for merchants to carry out targeted marketing. Smart merchandise guide system through interaction with user, analysis their behavior, their preferences and recommending the merchandise that they really interested in, and it is similar to the function of purchasing assistants to help users to filter information, to purchase the merchandise to meet their real needs. Smart merchandise guide system can also greatly enhance shopping experience and provide personalized services for user, it also can promote product sales and improve user loyalty in the fierce competitive environment for e-commerce enterprises.At present,majority of B2C e-commerce systems exist merchandise recommend strategy single, recommend results's accuracy lower, and be shot of personalized questions. In this paper, we research on non-linear network user behavior, modeling what users interested used an improved non-linear regression analysis method, using AJAX technology to finish the data source collecting that user interested in, replace the complicated process of web log,and analysis advantages and disadvantages of existing recommend strategy, specially on the association rules and collaborative filtering recommendation algorithm principle and so on, introducing a personalized recommend algorithm based on the merchandise characteristics and genetic algorithms, effectively improved the coverage rate and accuracy of merchandise recommend.we introduced the concept of the characteristics of merchandise, by combining the database of the merchandise characteristics, the history user purchasing, the content user browsing online and the user behavior in neighboring, proposed a personalized recommend algorithm which is based on characteristics of commodity and genetic algorithms. this algorithm can make up for recommend low accuracy, low efficiency, not recommend timely and so on, at the same time studying candidate sets which customers interested, obtain the best neighbor-user mode to upgrade merchandise recommend coverage rate.Finally,we set up an open recommendation system prototype based on the J2EE framework.