Research And Application Of The Social Network And Collaborative Filtering Recommendation Algorithm

With the rapid development of social network, internet users are facing terrible information overload problem. In order to alleviate “choose difficult disease” problem incurred by mass data, the personalized recommendation system is widely applied to the major social network and e-commerce platforms successively. However, as the recommendation system is widely used, the traditional recommendation algorithms cannot meet the demand of the current platforms. Therefore, it is very meaningful to study recommendation strategy to reduce the pressure from big data, and improve the comprehensive quality of recommendation system.Main research contents and application value of the thesis are organized as follows:Based on existing researches, the thesis proposes a twin-engine recommendation system framework for big data by analyzing the present status of recommendation technology on the parallel platforms. In addition, a twin-engine recommendation system framework which is for big data with the particular scenario is designed according to the characteristics of algorithm, such as the iteration and complexity, and groups of contrast tests to analyze different specific performance of specific algorithm. The recommendation engine is on demand in the form of component assembly into the framework according to the dynamic demand for single or distributed recommendation service. And the effectiveness and practicability of the proposed framework are showed by a large number of contrast and analysis examples.Based on the above twin-engine recommendation system framework for big data, study the defects in the collaborative filtering recommendation algorithm. It is possible to solve the problem of single user similarity calculation method to search the availability of social network data and introduce the social network user relationship as the calculation basis. And then the collaborative filtering recommendation algorithm with social network properties is proposed to locate the relatively accurate neighbors for target user and improve the accuracy of the recommendation system. The experimental results indicate that the proposed algorithm is more accurate and personalized than the traditional methods on the aspect of similarity calculation by using the social network elements effectively. The algorithm application enriches the result interpretation of recommendation system and improves the comprehensive quality of the recommendation system.In conclusion, the thesis designs a twin-engine recommendation system framework for big data, and within the framework proposes a social-based collaborative filtering recommendation algorithm through a lot of active exploration and research, finally the new designed and developed algorithm application runs well, and which greatly enriches theory research of the collaborative filtering recommendation algorithm based on social network.