Hot Topic Discovery And The Application Of Word Cloud Based On Voronoi

Weibo has accumulated a large number of users due to the immediacy of its content dissemination speed and the wide range of dissemination,and has become the public opinion explosion and dissemination place of hot events.For a hot event in Weibo,a large amount of social media text data will be generated within a short period of time,and this hot event often contains several semantically related subtopics.Discovering various subtopics contained in the Weibo is of great significance for the correct guidance of public opinion.However,due to the semantic sparsity of short text,the effect of text mining algorithms such as topic mining or text clustering will be affected.In addition,the results of social media text mining are often presented through text visualization technology but the traditional text visualization model represented by word cloud has problems such as the scattered arrangement of displayed word items.Displaying the subtopic information contained in the text data by word cloud seems a little difficult.The main research contents of this paper are as follows:1.This paper proposes a text feature self-expanding algorithm based on the corpus itself.The algorithm believes that a certain word in the document is generated by sampling with a certain probability by the topic.Continue this sampling process to fill a number of dummy words into the original document will realize text feature expansion.Text feature expansion solves the semantic sparsity problem of social media short texts and improves the effect of text clustering algorithms,avoiding the problem of traditional text feature expansion algorithms relying on external data sources.2.This paper proposes a text visualization technology named VoronoiTopicCloud,which integrates the Voronoi diagram into the word cloud diagram,divides the plane into several regions through Voronoi,and aggregates semantically related terms together,which solves the problem of using word cloud as the representative text visualization technology cannot display topic semantic information well because of the scattered arrangement of the displayed terms and no surrounding context information.3.This paper designs and implements a microblog hot topic mining system.The system crawls microblogs through web crawlers and conducts topic mining.It uses the text feature self-expanding algorithm proposed in this paper to solve the problem of short text semantic sparsity,and finally uses VoronoiTopicCloud to present visual results,and realizes the public opinion discovery and public opinion tracking of hot topics on Weibo.The paper concludes with a summary of the entire research work and clarifies the future research direction at last.