Data Analysis And Personalized Recommendation Technology Research

The rapid development of Internet has brought the information overload problem, finding the content that people want from the vast amounts of information is becoming more and more difficult. In the face of the Internet era of information explosion, the personalized recommendation technology has become a key technology to solve the problem of today’s information overload.Recommendation technology analyzes the interest model from the user’s history information. Then, based on the user’s interest model, the recommendation system can choose the appropriate items to recommend to the user. This initiative behavior that recommendation system selects items from huge amounts of data for the user, can greatly improve the users’experience, so the study of recommendation technology has a certain practical significance.This paper first introduces the research and application status in the field of personalized recommendation, then introduces the theory of several important recommendation algorithms in detail, including the advantages and disadvantages of them. This paper mainly analyzes and researches Slope One algorithm. It is a kind of collaborative filtering algorithm based on least square method. Firstly this paper analyzes the problem that Slope One still faces, and by combining the tradition recommendation algorithm based on content, this paper proposes an improved algorithm. Secondly, after analyzing the parallelization feasibility of the proposed algorithm, by combining with the distribute platform Storm, this paper proposes a parallel algorithm with the storm framework. Experiments on dataset that have different size show the different time cost by training linear model between the serial algorithm and parallel algorithm. And the contrast test result is given. At last, the paper presents a movie recommendation system based on the proposed algorithm. And this recommendation system provides some main functions, such as collecting users’rating information and recommending the items that confirm to the user’s interests to the active users, etc. And through the movie recommendation system, the effectiveness of the proposed algorithm is also verified.