Design And Implementation Of Weibo Public Opinion Monitoring System

Network public opinion, under the impetus of the new network media such as weibo, has gained unprecedented communication speed and social influence. Through the weibo public opinion monitoring, the relevant functional departments of the government can track dynamic social affairs, network hot spot of public opinion, and to identify and guide various topics on the Internet, so as to purify the network environment, and improve the social atmosphere. The relevant functional departments of the government also can grasp public sentiment, public opinion, to provide a basis for government decision-making.This paper studies the related technology of network public opinion monitoring based on Sina Wei Bo, and designed and implementated a weibo public opinion monitoring system based on B/S architecture. Its main functions include: the data acquisition of weibo, the weibo clustering, weibo cluster summary, and timeline segmented display. The main tasks of this paper include:(1) The weibo crawling. The carwler of this system is based on the user-defined keywords to crawle weibo search results pages, so as to obtain the specific weibo data, and reduce the response time, as well as the hardware costs of weibo data storage.(2) The weibo clustering. This paper uses the modified Single- Pass clustering algorithm based on center vector to cluster similar weibos, combined with the weibo hashtag, link url and weibo user name, to improve the effect of weibo clustering.(3) The weibo cluster summary. This paper uses a improved hybrid TF-IDF algorithm to calculate the keywords and ratings of weibo cluster. Then, through the keyword feature vector, the most similar weibo is selected as the summary of weibo cluster.(4) The timeline segmented display of cluster summary. This paper uses the Chart Director chart plug-in to segmented display weibo cluster summary, in order to achieve the purpose of monitoring, analysis and tracking of weibo public opinion.Finally, this paper uses the Tweet Event2012 data set to evaluate the related clustering and summary algorithms.