Research And Implementtation Of Micro_blog Recommendation Algorithm Basd On Content And User Behavior

With the development of information network,micro-blog become the biggest social information acquisition in China.With the exponential growth of users' information,the method of getting information that they are interested in is a hot spot of research.The recommendation of micro-blog is mainly based on users' social network and micro-blog content,which ignores the needs of user and micro-blog's characteristics,So it can't make micro-blog recommendation precise.At the same time,the micro-blog recommendation is based on the entire social network,which needs a lot of time.According to these problem,paper put forward personalized micro-blog recommendation algorithm based on content and theme preferences.Micro-blog is subdivided in order to improve the accuracy of the subject.Then,the community can reduce time.The main research content includes the following three points:(1)The forwarding micro-blog and original micro-blog will be treated separately and the comments of forwarding micro-blog is used to sentiment analysis,correct the topic preference of forwarding micro-blog based on the result of emotion analysis.The weight fusion was made to obtain the topic probabilistic model according to the original micro-blog and the modified forwarding micro-blog,finally the paper present a method of micro-blog recommendation based on sentiment classifications combining collaborative filtering.(2)This paper developed the definition of micro-blog quality and the definition of active users,based on the characteristic that active users pay more attention to micro-blog quality,we recommend micro-blog to active users based on micro-blog content,based on the characteristic that inactive users pay more attention to micro-blog publisher,we recommend micro-blog to inactive users based on social relationships.We developed a personalized method for micro-blog recommendation base on micro-blog content.(3)This paper divides community according to the social network and theme preference,and recommends users in the community to reduce the complexity of real-time recommendation.Recommendation based on community can solve the problem of cold start for paid users.(4)In this paper,three methods are combined to implement a micro-blog recommendation system.The system combines the advantages of the three micro-blog recommendation methods,which realizes micro-blog's real-time recommendation with high accuracy and low consumption.