Research On Microblog Recommendationmethod Based On User Social Behavior

With the rapid development of Web2.0 technology and mobile terminal equipment,the global Internet has entered the network interactive age.As typical representative of Web2.0 technology,microblog attracts many users thanks to its convenient and efficient,dissemination and strong interactivity.It has gradually become an information exchange platform for information sharing and opinion expression.The explosive growth of microblog information also caused the proliferation of information.Therefore,how to excavate the content of user's interest and accurately recommend appropiate microblog to the target user from massive microblog data has become the focus of microblog research.The decision process to recommend suitable microblog posts for users is complex and can be influenced by numerous factors.A new microblog recommendation framework based on the integration of microblog content,user tags and user social relationships is presented via analyzing microblog features and the deficiencies of existing microblog recommendation algorithms.The framework includes three parts: microblog user tag expansion,micro-blog user interest modeling and microblog recommendation.The purpose of the microblog tag expansion is to recommend the high quality microblogs for users by extracting the keyword in the micro-blog text as the user's tag.The purpose of microblog user interest modeling is to make the system "do the right thing in the right time",that is,to give users the best experience.In summary,the contributions of this paper are as follows:(1)A microblog tag expansion scheme based on hypergraph random walk is proposed.We model posts and term associations from each user as a hypergraph.The weighting strategies for both hyperedges and hypervertexs are established,which is able to model post temporal and social features,as well as correlation weights for term features.Random walk is then performed on the weighted hypergraph to acquire the higher ranked terms as expanded tags.(2)A microblog user interest representation model based on the tag probability correlation is designed.Firstly,we take advantage of the probability correlation between tags to construct the tag similarity matrix.Then the weight of the tag for each user is enhanced based on the relevance weighting scheme and the user tag matrix can be constructed.The matrix is updated using the tag similarity matrix,which solves the sparse problem of user tag matrix.(3)A microblog recommendation method which integrates the user social relationships is presented.On the basis of the previous work,we provide an iterative approach for updating user-tag matrix via the user social similarity matrix based on co-follower and co-followee to reveal the user's interests more accurately.We report extensive experimental evaluations to demonstrate the superiority of the proposed approach over the existing recommendation methods.