The Research Of Recommendation Algorithm Of The Social Network

The Internet has speed the flow of information on the Internet, thus bring a sharp increase in the amount of information, which has led the users’ selection for information became extremely difficult. Usually, people filter information through search engines, etc. After the arrival of a web 2.0, users’ interaction began to start, people came to be unsatisfied with the search results from the search engine, from that time on, the recommendation system formally confront to people. Recommendation system provides users more personalized recommendation through a user’s past information, which could be widely used in e-economy, etc. this can bring huge economic income and attract a large number of experts involved in the research. In recent years, big data and the arrival of the Internet plus have people the information dimension of ascension, a large number of industry has becoming visualizing based on society, the data collection from all walks of life become easy, usually the data can be gain through We Chat and other social software’s connection, making recommendation of social network has become very meaningful. Similarly, in the existing recommender systems, the using of more considered recommendations from the local data or multiple sources, multiple dimensions simultaneously recommended system is not common. as a kind of special social network, Q&A community has a problem that the answer, the more elements of users, such as users’ behavior in community is more complex, the information dimension is quite high. Data dimension in the future will continue to improve as the trend of the research become more meaningful.In this dissertation, by combining the characteristics of the Q&A community access to information, the research and theoretical study of the basic algorithm of the recommendation system of knowledge, expounds the influence of social networks in star, gives the star the opportunities to find the related theory; Through the theoretical basis for q&a community questions and answers, subject areas, user modeling analysis respectively, to the user’s behavior connect each part to form the overall model; Through model applying collaborative recommendation to the user- answer- question- the problem under the recommended topic structure, through the answer, the question and the topic of transformation of recommendation system sparse problem; Based on the analysis of the content in question and answer community star found that found under each topic area influential celebrities, by topic(topic tree), the influence of relationship between correct topic under the stars; Through the star data, response time, the attention of the relationship between the user on the answer to the problem under the factors such as recommended, by exposure times to realize the dynamic of the answers is recommended; Dynamic structure definition, through the way of fragmentation of dynamic data segmentation, split the data for dynamic recommendation, in order to solve the problem of cold start; Recommend essential question, calculate the overall data found in recommended essence problem would solve the problem of cold start.