Font Size: a A A

Social Network Public Opinion Analysis Based On The Microblog Platform

Posted on:2016-07-03Degree:MasterType:Thesis
Country:ChinaCandidate:S Q HuangFull Text:PDF
GTID:2308330473965493Subject:Computer technology
Abstract/Summary:PDF Full Text Request
In recent years, microblog has become a kind of national network media form with rapid development and great influence. It becomes convenient for netizens to comment on various hot topics, which can usually make response ahead of the traditional media to make the effect of the event detection. Thus, deeply digging the themes behind amounts of inter-text objectives, analyzing the inner semantic associations and studying its evolutionary trend not only have great theoretical research value, but also contain huge economic and social value.The traditional topic detection method is mainly used to traditional media text while its processing effect on short microblog essays that are sparse. In this paper, we concentrate on the rich inter-text context-oriented information existing in the microblog data. Firstly we puts forward the pretreatment method of data weight based on microblog’s structural figures. Then, in account of the problem of short texts’ sparse characteristics, it uses the clustering algorithm on the basis of term vectors, in order to enrich the data’s characteristics information. On this basis, it extends the traditional and potential LDA topic model. Compared with traditional algorithms, the method proposed in this paper has better performance in Sina microblog test corpus, and its effect is better than those of the single clustering algorithm and topic detection model.Based on the selection of topics, the evaluation model of microblog information’s spreading influence based on PageRank is also put forward in this paper. This model pays more attention to the influence of the microblog data’s structural features in the spreading process to the quality of data’s spreading to summarize the three aspects that affecting microblog users’ behavior. Finally, it adopts the PageRank algorithm for the calculation, and by contrast, the result that the method proposed in this paper is obviously better than the traditional PageRank in evaluating microblog’s influence.Finally, by analysis and research to the specific cases, we design and Implement the public opinion analysis prototype system, the feasibility of the topic detection method and influence evaluation model is proved in view of practice.
Keywords/Search Tags:Microblog public opinion, Topic detection, LDA, Spread influence, PageRank
PDF Full Text Request
Related items