Personalized Recommendation Techniques And Its Applications For E-Commerce

With the speedy development of internet and E-commerce techniques, peoples urgently ask for a personalized recommendation technique that can help them realize the information filtering and automatic recommendations of the product. This paper researches the personalized recommendation system and techniques especially the collaborative filtering technique including user-based and item-based.Collaborative filtering is the most successful technique of personalized recommendation techniques for building recommendation systems and has shown special superiority for recommendation effect and accuracy, but its performance is seriously suffered from the problems of cold-start and sparsity, which makes its effect can't exert completely.To solute these issue, we useed an algorithm combining item-based recommendation and user-based collaborative filtering recommendation together to make up the limitation of user-based collaborative filtering recommendation for recommendation of the new items by research on the similar items in item-based recommendation. The combining algorithm analyzes the characteristics of attributes of items and finds the similar items to the new item by item-based recommendation firstly. Then, it forecasts the rate to the new item from user by the degree of these similar items and the rate to these similar items from user. Finally, it computes the neighbor users on the similar items by collaborative filtering recommendation and presents the final rate of forecast, so that the new item can participate in the process of recommendation, and sparsity is solved at the same time.In this paper, we introduce the component and structure of personalized recommendation systems and primary recommendation techniques, analyzes the root of the problems of collaborative filtering recommendation. Finally,the application of the personalized recomender systems with the hybrid algorithm in the industry of electric supermarket is introduced.