Research On Electronic Commerce Recommendation System Based On The Collaborative Filtering Algorithm

Along with the development of e-commerce and the "information overload" problem, how electricity enterprises satisfy consumers’demands for accurate and personalized information, enable users get goods or information in the vast amounts of information at low cost, enhance consumers’experience and satisfaction, is especially important for traditional electrical contractor. With the development of mobile Internet, recommendation algorithm is faced with new challenges, mobile electricity for its accuracy and targeted put forward higher request, under the new situation, information recommendation becomes more specialized and complex.There are two basic recommendation technology:content-based recommendation and collaborative filtering recommendation. Both of them have advantages and disadvantages:content-based recommendation works with the content of the perspective of project, through the analysis of user access to the project characteristic, which would be similar to the projects of users never visited recommends to the user, its shortcoming is to ignore the potential interest of the user; In contrast, collaborative filtering, from the perspective of the user, through the analysis of the similarity of users, who would be similar to the user’s favorite project recommend to users, but the collaborative filtering has data sparse, cold start and other issues. Personalized recommendation system recommends goods that users might be interested in, which to a certain extent, relieve the information overload problem, help users to find information that meets the demands of their personality. In the mobile Internet environment, the terminal from the PC to a more personalized mobile phone, etc., on the recommendation of personalized and precision are put forward higher request. With the widespread popularity of the mobile terminal and mobile Internet technology advances, the way consumers get information has changed from traditional to mobile phones or tablet PC and other mobile terminals.This paper is based on these two technologies recommended principle, and on the basis of detailed analysis, summarized the advantages, proposes the model of user preferences based on content, when the user and the score under the circumstances of less data, achieve more accurate recommendations.This article uses the Netflix data sets and Taste recommendation engine for model’s validation. Finally this article will apply recommend system to the mobile phone electronic coupon software, then realize the coupons’ recommendation.