Design And Implementation Of Microblog Public Opinion Monitoring System

Social network platforms have gradually become a new medium of information communication between users.When something happens,users use the social network platform to post text,pictures and other information to express the attitude about it.This information is further spread through social networks,which has a certain social impact.On the one hand,public opinions counld be conducive to the government and management departments to understand the people’s sentiments and serve the people better;on the other hand,public opinions may also be used by criminals to incite mass emotions and undermine social stability.In this context,the importance of public opinion monitoring systems is self-evident.At present,Sina Weibo in China can reflect the public’s grievance to a large extent.Therefore,monitoring Weibo public opinion information has important practical significance.This paper analyzes the research status of public opinion monitoring and the operation of its application system at home and abroad,elaborates the demand analysis of Weibo public opinion monitoring system,and designs and develops a microblog public opinion monitoring system by using relevant models of deep learning.The system has functions such as data acquisition,data preprocessing,data macro analysis,event tracking,hot event prediction and negative public opinion monitoring.The core function modules and principles in the system are as follows:1.Event tracking module.The module uses the Bi-GRU model and the Attention mechanism to construct an event tracking model.Considering the dynamic increase of the information cascade size over time,the cascade structure information of different time intervals is intercepted as the input of the event tracking model.Predict the spread of the same information in different time intervals in the future;2.Hotspot event prediction module.Because the event heat is often positively correlated with its cascading scale,based on this phenomenon,this module embeds the information cascade structure to predict the number of information forwarding based on the event tracking model,and judges the hot event based on the predicted times of retweeting;3.Negative sentiment monitoring module.The module performs jieba word segmentation on the microblog text published by the user,establishes a Bow document dictionary,and then builds a deep emotion classifier based on the Bi-LSTM model to realize the detection function of negative lyrics.The simulation experiment results on Sina Weibo data show that the public opinion monitoring system can effectively predict the public opinion trend of Weibo data and make accurate predictions for future hot events.The system can be used to collect public opinion in the future,and guide the government and management departments to improve decision-making and maintain social stability.