Research And Application Of Oriented Social Network Public Opinion Capture Analysis Strategy

In recent years,with the continuous development of Internet application technology,the number of users on the social network is increasing,which makes the Internet public opinion become one of the main sources of opinion data in today's society.Network public opinion is collected attitude,comments and views from Internet forums,microblogging,WeChat and other social platforms with a large number of users on hot spots of social events,which has an important impact on stability of society and the healthy development of the Internet.Timely capturing the social hot spotsd and correct analysis of public opinion has become an important research topic today.The traditional technologies of network public opinion analysis are based on open source technology,and open source technology is wide but not deep,and it can't fully take into account the special data format in the social network.Cross-domain data also affects the accuracy of public opinion analysis.Furthermore,the coupling between each technical modules is too high,which will make the whole public opinion system maintenance and scalability worse.This thesis is devoted to the research of public opinion capturing and analysis for social networks.The main research contents include Chinese word segmentation,text clustering and emotional tendency analysis.According to the characteristics of fast and transcendental information in social network,a simple,efficient,extensible and easy algorithm to maintain Chinese word segmentation structure is proposed,and a Chinese word segmentation algorithm is presented to improve the processing speed of Chinese word segmentation and accuracy.This thesis presents a k-means text clustering algorithm based on the maximum density,designs an initial point selection strategy,to solve the shortcomings of k-means algorithm whose clustering results are unstable and which is sensitive to noise points,timely identify noise points,reduce invalid iterations and ensure the correctness of the clustering results.This thesis designs a multi-dictionary and multi-rules text emotion tendency analysis algorithm.According to the characteristics of colloquialization,randomization and diversification in social network,it constructs the emotional dictionary with good performance,high fault tolerance and wide information coverage,and analyzes the text data sentence structure,with a comprehensive consideration of the various situations appear in the sentence and the design of the corresponding rules for emotional score calculation.Finally,this thesis adopts B/S architecture to build a network public opinion visualization platform,where the proposed algorithms and data structure were verified and analyzed.The results of clustering and textual emotional tendencies are presented in graphical form.The results of the experimental study show that the proposed algorithms can analyze the network public opinion effectively,and provide a powerful force for the government departments to respond quickly to the change of the public opinion.