Research On Personalized Recommendation Using Multi-interest Model And Social Network

With the development of information technology,E-commerce enters a high-growth era.However,flooded with huge amount of products,customers have to spend hours filtering irrelevant information by themselves which is often called as “information overload” problem.The recommender systems are designed to handle this problem by analyzing the preference of customers and pushing products that they may like to them.To make precise recommendations,the recommender system should capture users' various interests and build the user interest model.The system can recommend the right products based on the interest model.Besides,it is another hot topic of current researches that using social network to enhance the effect of recommendation.In this thesis,a literature review on personalized recommendation is given at first.Then a multi-interest model is proposed and combined with the user-based collaborative filtering(CF)algorithm.Besides,the social network information is used and the interests of social friends are introduced to improve the multi-interest model.Experimental results show that the proposed algorithm can achieve higher recommendation performance compared with some conventional CF methods and to a certain extent,can handle the new user cold-start problem.