Research On Recommended Method Of Microblog Query Based On User Interest

Microblog as a representative of social networking products,to provide people with a fair and open real-time communication platform,every day users can directly participate in the message manufacturing and dissemination process,making microblog daily mass information,how to microblog data from the massive Obtain the information to meet the needs of users is microblog query recommended to solve the problem.As the microblog text is short,including complex information,the traditional query recommended technology and can not be well applied to microblog query recommendations.In order to solve this problem,this paper takes Sina microblogas an example to study the recommended methods of microblog query,and focuses on two problems in the recommended method of microblog query:user interest partitioning and query recommendation relevance correlation algorithm.In order to solve the problem of user interest,this paper improves the LDA theme model and obtains a user theme model based on forwarding relation.First,the use of microblog between the relationship between the forward microblog will be forwarded as the original microblog expansion of the text;Secondly,in the LDA theme model based on the introduction of the author layer,the LDA original document-theme distribution is divided into documents-author And the author-subject distribution;then,according to the matrix algorithm,the author-subject and the subject word matrix are used to calculate the set of words and correlations of each author,which is the user's interest label set.Finally,Based on the use of cosine similarity calculation method to calculate the similarity between the two users.Aiming at the problem of relevance ranking of query recommendation results,this paper combines the microblogging publishing time characteristics and microblogging structure characteristics and user authority three kinds of characteristics,and proposes a method of ranking the relevance of query results.This method analyzes the characteristics of publishing time and microblog,which are commonly used in the existing query relevance ranking method,and increases the authoritative characteristics of microblog users,which are represented by user's interest.Combining the 13 sub-features of the three types of features,And the optimal feature set is selected by the optimal feature selection algorithm.The optimal feature is sorted by the sorting feature vector,and finally the query recommendation method is satisfied.Through the experiment and test on the Sina microblog dataset,the feasibility and effectiveness of the scheme are verified.The experimental results show that the proposed method has improved the average accuracy rate compared with the traditional query recommendation method.