Research On Hot Topic Identification Of Network Public Opinion Based On Community Discovery

With the development of WEB2.0,the network has become an important channel for the rapid aggregation and dissemination of many hot topics,how to identify hot topics on network platform effectively is a hot research direction in the field of the current data mining.However,grassroots users,convenience using,timeliness of information dissemination,interactive mode and unique content of fragmentation of information makes the topic is easy to spread,to the national security and social stability brought great hidden trouble.So,identify and mining hot topic accurately and efficiently from the massive network public opinion information has great practical and scientific significance,not only can help the user to timely access to accurate hot topic information,but also can assist the government departments to control the development trend of the hot topics in real time,reasonably control and guide the direction of public opinion.Network information mainly includes two aspects:media platform and user platform,media platform mainly for news sites,audio and video sites,search sites,etc;user platform mainly for micro-blog,blog,e-mail and so on,no matter what kind of platform will be flooded with spam information data,it all has brought great distress to the identification of hot topic.In this paper,by analyzing the content,heat and emotional tendency of network information,combining with the characteristics of the data,the paper makes a deep research on the identification of hot topics and the migration of hot topics.First consider the basic structure and characteristics of the two text platform,data preprocessing,including word segmentation,noise filtering,cleaning and so on;secondly,Secondly,from three aspects of dimensions,characteristics and measures,construct topic index system,establish the second network model of "post-topic" furthermore,research on keyword extraction based on the effective keywords required by the model,choose the word frequency,word frequency growth rate and topic weight three features to filter key words;then,select community discovery algorithm in the aggregation method to do the topic community discovery,set keyword for node,co-occurrence keywords for edge;finally,through the user influence and the spreading influence of the dual measure to establish the topic heat to identify the hot topic,at the same time,do the topic heat transfer and the distribution of emotional tendency.