Research On Topic Recommendation Based On User Clustering In Micro-blog

Over the past decade,the rapid development of the Internet,the popularity of the Internet and the size of China's Internet users has increased dramatically.With the development of the Internet,micro-blog as a platform for Internet users to share and publish information has also been widely used.Micro-blog scale has also brought an explosion of growth on the topic of the micro-blog platform,and there is a certain degree of contradiction between the large number of topics in the micro-blog platform resources and user identification capabilities.Faced with the needs of users to browse and pay attention to the topic,micro-blog platform is difficult to provide a large number of topics in the topic to meet the user's preferences.Based on the above research background,this paper puts forward recommendation system to recommend micro-blog topic to user,the system can be effective in micro-blog's vast resources to find out the topic meet user preferences topic collections.Reduce the time cost of micro-blog users to browse the topic,improve user efficiency.Specific research contents are as follows:First,the calculation of user influence in micro-blog.The high impact of the user's attention to the topic of micro-blog is often easier for others to pay attention(such as a star of the topic is very easy to be concerned about their fans).In micro-blog,user often follows other user,in this paper we use the user network diagram to analyze and save the user relationship,and on this basis,using an algorithm(similar to PageRank)is calculated for each user's influence.Secondly,the expression form of micro-blog user preference.How to express the topic preference of a user is a difficult topic in micro-blog.In this paper,we Chinese according to word segmentation,feature extraction and mining association rules,the same topic of micro-blog and micro-blog based on user history,put forward the topic feature map(TFG)and users of micro-blog feature map(UTFG).By using TFG and UMFG,the user topic feature matrix(UTFM)are abstracted to represent the user's preference.Finally,accurately and efficiently recommend topic to user.Users of the same preference tend to love the same topic.In this paper,based on the UTFM,we use the cosine distance to cluster the users through the agglomerative hierarchical algorithm.In the same cluster,we recommend the user with the heat of the topic and the influence of the user.Through the analysis and verification of the recommendation system,the recommendation system in the calculation of user influence and define user preference topic expression and recommended micro-blog topic algorithm satisfies certain rationality,can accurately and efficiently to meet user preference topics recommended to the user.In the case of a huge number of micro-blog topics,can reduce the user's cognitive burden,greatly reducing the time required for users to browse micro-blog,improve user efficiency.