Test Analysis And Visualization Of Earthquake Public Opinion Based On Weibo Data

Earthquakes are one of the most serious natural disasters that cause damage,and major earthquakes can cause serious losses in terms of casualties,resulting in economic damage and destruction of infrastructure.The traditional way of obtaining disaster information is characterized by long time period,large workload and lagging behind the disaster situation,which cannot meet the needs of the government to carry out real-time relief work.Sina Weibo,a social media platform,is recognized by the public for its ability to post information about what is happening around them anytime and anywhere,and it is interactive because it can post text,pictures,and videos,as well as leave messages and respond to visitors’ comments anytime and anywhere.The suddenness and destructiveness of earthquake events and the freedom of social media platforms overlap to form online public opinion,which drives the course of events to a certain extent,and can make up for the shortcomings of traditional disaster relief by fully exploring the public sentiment and hot topics reflected in online public opinion.It can provide important reference for disaster research and emergency management.The thesis first uses a Python tool to obtain Sina Weibo data,takes Yunnan Yangbi earthquake and Qinghai Mado earthquake as the research objects,analyzes the spatial and temporal evolution of earthquake public opinion features through Weibo index and thematic maps,and uses deep learning-convolutional neural network(CNN)and implicit Dirichlet distribution(LDA)model We extract finer-grained topic information to obtain public attention hot topics,construct a network association model using Louvain algorithm to mine the structure of hot topic associations,and finally use sentiment mining techniques to study the spatio-temporal evolution characteristics of microblog users’ sentiment tendencies to obtain two earthquake response results.Based on the main research content of this article,the following conclusions are drawn.(1)Public opinion data collection and processing.We use Python to collect microblog data,clean,de-duplicate and remove microblog data that are not related to the earthquake,and call the stuttering word splitting tool in Python to split the text.(2)Study on the heat of earthquake public opinion.According to the statistics of public Weibo posts during each time period,the discussion trend corresponding to the Yangbi earthquake and the Maduo earthquake is positively correlated with the evolution of the popularity of original creators.By counting the number of Weibo posts in each province,the conclusion is drawn that the epicenter and high-density population areas have the highest level of public opinion response.The high-frequency words of earthquake public opinion were obtained by Chinese word separation and word frequency statistics,the heat of public opinion was visualized in the form of word cloud map.(3)Earthquake public opinion information extraction method.For the shortcomings of traditional opinion information analysis,a method based on deep learning algorithm is proposed to extract earthquake public opinion information,using convolutional neural network method to extract information on Yangbi earthquake topics and LDA topic model to extract hot topics of Mado earthquake events,concluding that different city/county regions pay different attentions to earthquake topics and the closer to the epicenter the higher the degree of public response to earthquake events.The Louvain algorithm of Gephi software is used to construct the network association model of public opinion hot topics and to divide the associations.(4)Public opinion sentiment mining techniques.An emotion dictionary was constructed to determine the public’s emotional tendency,and it was concluded that the network attention of both earthquakes was high and the microblog public opinion was dominated by positive emotions.(5)Analysis of the temporal and spatial changes of online earthquake public opinion and the driving forces.Focusing on the correlation between public opinion hotness,hot topics,public sentiment and earthquake development in both time and space,the final analysis of driving forces for the evolution of public opinion on the two earthquakes found that traffic factors,aftershock factors would affect the evolution of public opinion,indicating that the evolution of public opinion is related to the development nodes of earthquake events...