Research On The Collaborative Filtering Recommendation Method In MC Based On User Interests

With the wireless network covering a large area, the rapid development of ecommerce and the rapid adoption of smart phones, make Mobile e-commerce(MC) grow rapidly around the world, its personalized recommendation system has become the tool which can improve business competitiveness, meet the needs of the individual users. However, the particularity of the mobile e-commerce make the traditional recommendation system hardly simply transplant and meet the special needs of the age of "digital universe". Mobile e-commerce is a mobile commerce service with a high degree of situational dependence, when the user is personalized recommended, the user features, the project features and current characteristics of the user will influence the user's interest.In order to meet the individual needs of mobile e-commerce, improve the quality of recommendations. In this paper, based on the consideration of the influence factors of mobile e-commerce users' interest, built the three dimensions of users' interest model which based on user feature dimensions, project characteristics dimensions and situational characteristics dimensions, designed MC collaborative filtering methods based on user interest through the process of collaborative filtering. Firstly, using the method of weighted Slope One to fill rating data for mobile e-commerce user interested model to solve the sparsity problem of score data in the process of collaborative filtering recommendation.Secondly, cluster of mobile e-commerce users through firefly improved method based on K-means clustering, improving the accuracy of the selected cluster centers, reducing the target user's nearest neighbor search space. Finally,improved the similarity calculation collaborative filtering method by introducing a weighted multidimensional similarity theory, so the similarity measure andrecommendation become more comprehensive and more accurate, provide more personalized and more accurate recommendation results for mobile e-commerce users.