Research On Representation And Evolution For Micro-blog Hot Topic

Multi-subcenters will be generated by the initial topic center during the discussion of the micro-blog hot topic.The topic content is difficult to be understood timely and comprehensively because of micro-blog characteristics such as low value density,fast update speed etc.To solve this problem,This paper proposes a Representation and Evolution Method(REM)for Micro-blog Hot Topic to help users understand the topic content better.First,micro-blog hot topic keywords are extracted and co-occurrence matrix is created.Then,topic complex network is constructed to express and analyze topic content conveniently.The main research contents are as follows:(1)Constructing micro-blog hot topic keywords co-occurrence matrix to show keywords and their correlation strength.First,the micro-blog hot topic dataset are obtained by using the "Seimicrawler" crawler tool.The "Jieba" Chinese word segmentation tool is used to segment the Chinese text.Then,the combination of the micro-blog importance(e.g.,the number of comments and the number of praises)and the time decay coefficient are made to calculate word support.Formula of demarcating the high-and low-frequency terms is used to extract the words with the highest ranking of support.The strength of the association between keywords is measured by the complex co-occurrence rate of words.Finally,the keywords co-occurrence matrix is created.(2)Constructing micro-blog hot topic complex network to show the content of topics and the evolution of the hot topic core content.First,hot topics complex network is constructed based on co-occurrence matrix.An improved weighted modularity function is proposed to measure the quality of complex networks community division.The genetic algorithm is used to solve the maximum value of weighted modularity function to obtain the semantic community division result of complex network.Then,Scott's reference rules should be used to determine the size of time series.Finally,"Gephi" software is used to demonstrate evolution of the hot topic core content with the time coursing.The experimental results show that the evolution of micro-blog topics in different time series can be effectively visualized by REM.This paper innovatively uses multiple elements of micro-blog to extract keywords and co-occurrence words for the construction of complex network.The evolution of the topic over time is demonstrated by dividing different time series.It is the application of complex network theory and date data mining technology to the topic representation of micro-blog platform,which is extremely helpful to discover and understand micro-blog hot topics.Meanwhile,our research provides theoretical support for government to understand and control public opinions in social media.Figure[21]table[10]reference[62]...