Optimization Algorithm Of Retrieval Result In Microblog Based On User Personalization

In recent years,the rapid development of Internet technology and information technology makes it possible to communicate through the network.Among numerous social network platforms,microblog has become a social platform which has been widely used,the most popular and research scholars are concerned on.And in terms of real time,the microblog retrieval service has been approved.However,after a large number of research and analysis,we could find that microblog retrieval service does not do personalized search in today's people-oriented and the pursuit of personalized service,resulting in a waste of information resources of the rich microblogging users,and retrieval results are somewhat unsatisfactory.Studying a large number of retrieval engine technology,we acknowledge that the query expansion mechanism can effectively improve the accuracy of query.Based on the above findings,this thesis presents a new algorithm for the optimization of microblog retrieval results based on user personalized features.This algorithm assumes that the retrieval results of the microblog retrieval engine contains all the documents related to the query terms.Based on this assumption,the thesis uses Topic Model and weighted association rules to expand the semantic relevance of the query terms.Using the traditional similarity calculation method TF-IDF to get the similarity of query terms and documents,and adding the timeliness characteristic of the microblog platform,these two are all as the basis for re-ranking retrieval results,and then optimize the retrieval results.Making microblog user's information of post,forwarding and so on as the dataset,the experiments of the thesis is based on the results of microblog retrieval engine,and makes it as compared experiments that experiments of optimized results based on dictionary query expansion and traditional association rules.The experimental results show that the proposed algorithm,that is microblog retrieval result optimization algorithm based on user's personalized features,significantly improved precise compared to the benchmark results and the above two kinds of contrast experiments.