Hot Traffic Accident Information Acquisition And Public Opinion Analysis Based On Web Crawler Technology

With the rapid development of the network,the influence of network public opinion has attracted people’s attention.However,in the aspect of transportation,there are few studies on the influence of network public opinion,and it is at the initial research stage of traffic public opinion;especially for the network public opinion of traffic accidents.There is almost no research.However,improper handling of network public opinion on traffic accidents will cause many adverse social impacts.Therefore,studying the network public opinion of traffic accidents has important theoretical and practical significance.At present,the online public opinion text vocabulary in the field of traffic accidents borrows other mature vocabularies,and does not have its own exclusive text sentiment vocabulary,which reduces the accuracy of online public opinion in traffic accidents.This article establishes an online vocabulary of sentiment texts related to traffic hotspot accidents,and on this basis,calculates the ratio of positive and negative sentiments to the online public opinion,and finds out the control indicators for the outbreak of negative online sentiment.The thesis first studies the method of acquiring public opinion on the Internet and information pre-processing.Through the web crawler,public opinion information of hot traffic accidents is obtained;through text processing methods,preliminary analysis and text pre-processing are performed on the obtained network public opinion.The results show that: the preliminary analysis of network public opinion of hot traffic accidents is conducive to the control of public sentiment and emotion;the keywords of hot traffic accidents extracted by the text preprocessing of network public opinion have a certain representative effect on network public opinion.Secondly,the paper studies the construction of deep learning models based on convolutional neural networks and long-short-term neural networks.According to the combination of the two models,the network public opinion text is reduced into word vectors,which improves the memory time of the model.By dividing the overall network public opinion into training and test set training methods,a public opinion text vocabulary for hot traffic accidents is constructed.The results show that through the comparison of the precision and loss of the deep learning model,the public opinion text vocabulary is more in line with the expected requirements.Finally,through the use of three public opinion analysis methods,taking the traffic hotspot accident-Chongqing bus crash accident as an example,an online public opinion sentiment analysis was conducted to obtain the overall sentiment ratio of positive sentiment and negative sentiment of public sentiment;the corresponding control indicators for the outbreak of negative public opinion were proposed And strategies;and predicted the network public sentiment trend of this accident.The results show that the sentiment vocabulary of the network public opinion text using the established hot traffic accidents is more sensitive to predict the outbreak of negative emotions.The research results of this paper not only help to fill the vacancy of traffic accidents in network public opinion,but also help relevant departments to timely control the outbreak of negative public opinion caused by traffic hotspot accidents.