Research Of Mobile Terminal User-based Recommendation System

Because of The rapid development of the internet, people have changed their businesslife-style from the practical to the internet for the past few years.The electronic commerce developed very fast, more and more products is choosed by people in the internet.Accompanied by the production of this kind of situation, and by promoting e-commerce operators, now the development of electronic commerce in our life so fast that people on the Internet e-commerce behavior constantly incr-eased,more and more products in the Internet for people to choose, the common way of screening(search engine) has been difficult to meet people’s needs.With the rapid increase of smartphone users,people into a computer office applications,applicat-ion of the era of smartphones entertainment. Smartphone users increase will promote the develop-ment of intelligent software applications, also makes more and more electronic commerce and ente-rtainment platform will shift to the smartphone users,so the recommendation system based on mob-ile terminal users need to be further research.This era is an era of big data, how to go from vast amounts of data of exploding, quickly and efficiently find what they are interested in, is the rec-ommendation system based on intelligent terminal users need to face the problem.Based on the research status quo of traditional recommendation system as a starting point,thr-ough the study of traditional recommendation system and research on the characteristic of intelligent terminal, aiming at the particularity of intelligent terminal, according to the requirements of intelligent terminal users to improve the traditional recommendation algorithm, in the process of research on collaborative filtering algorithm, diversified algorithm is improved. In this paper, the main work is reflected in several aspects:(1) improve the accuracy of the cold start users: according to the characteristics of mobile users of mobile to traditional collaborative filtering algorithm is improved, and the mobile user scenarios change into consideration, the degree of similarity calculation program and make recommendations.(2) improve the recommendation diversity: will recommend the number, take into consideration of the mobile user scenarios, enhance the diversity of recommend variety.(3) improve system extensibility: application of graphs, and use the Hadoop architecture in order to improve the joining problem with a large number of users and products, improve the recommendation system scalability.