A Microblog Topic Search Algorithm Based On Semantic Analysis

Microblog is a social network platform to share real-time information through short attention mechanism,The characteristics of large amount of real-time information,fast spreading,fragmented content.make it attracting many people’s attention and used widely,thus,it have a significant impact on the way of communication and spreading information.vast consumer base and huge amount of information make microblog to become a vital source of real-time information,many hot events were reported firstly by it.Microblog search users prefer to search recent hot topic or interested topic So studying microblog topic search is necessary to help user to get interested topic accurately in massive information.At present,microblog search is based on keyword matching,without analysising user’s search intention semantically,the search results is not good enough to meet the user’s expectations,so this paper proposes a microblog topic search algorithm based on semantic analysis,research contents include:firstly,to solve the problem that the cosine similarity cannot accurately measure text inherent semantics,it proposes a combinated similarity calculation method which integrate sematic similarity and KL distance,provding the foundation for the following study.Secondly,it proposes query expansion strategy exploit dictionary and user’s feature words to achieve double semantic extension on the initial query words,thus addressing the problem of short query terms,making the expanded query words are more fit the user’s query intent.Then,clustering microblogs belonging to the same topic is helpful to achieve effient topic search,so it fully consider structural features information of microblog which makes microblog topic mined model mining microblog topic more accurately and identifying subtopics.Finally,it studys a specific microblog topic search algorithm,which analysis the correlation between expanded query document and each microblog topic to get related microblog topic,then further analysis the correlation between expanded query document and each microblog subtopic belonging the related microblog topic,eventually return search results to user by considering the similarity between expanded query and microblog document,microblog document heat,and microblog document released time.Using microblog dataset to design and implement experiment,and result shows that compared to traditional search methods,the precision and recall rates of microblog topic search algorithm query have improved to some extent,microblog topic search algorithm is more applicable to microblog topic search.