Research On Key Techniques Of Public Opinion Based On Social Network

With the in-depth development of the mobile Internet and the arrival of the 5G era,all kinds of social are in sustainable development,and the Internet platform presents pan socialization.The typical representative of social networks in China are WeChat and micro-blog,which play an important role in the public opinion monitoring and analysis system.Compared with microblogging platform public opinion monitoring technology is maturing,there is a few works in the research related to WeChat.Therefore,this paper focuses on key techniques of public opinion monitoring and analysis technology based on the WeChat platform.According to the requirements of public opinion system in the social network,the overall system framework is designed,the data acquisition problems of the official accounts in WeChat are solved,and furthermore the network public opinions are sentimentally analyzed in this paper.The WeChat data analysis model is proposed in this paper by combining the word2 vec model and the machine learning,which can effectively improve the classification accuracy,precision and recall rate.The main work of this paper includes:(1)The framework of the public opinion monitoring and analysis system in WeChat platform is designed,and the function modules are designed and implemented.This paper describes the process of system task execution,designs and implements the data acquisition function,data storage function,text sentiment analysis function and public opinion report display function.(2)With the combination of Python crawler and Hook technology,the automatic acquisition of official accounts data in WeChat platform is realized.And the data will be stored in the MySQL database.In the data acquisition function,the working principle and acquisition flow of each component are described in detail.The structure and communication mode of the mobile terminal components and server components are clarified.The information is stored after de-noising,which provides data support for public opinion analysis.(3)Based on the research of feature selection algorithm and classification algorithm,the word2 vec model is further studied in this paper,and a WeChat text sentiment analysis model based on machine learning method and word2 vec is proposed.Compared with the traditional classification model,the optimized model proposed in this article shows a better performance in the WeChat text sentiment classification.The experimental results show that the model proposed in this paper performs well in the text sentiment analysis of WeChat platform.The accuracy rate reaches 84%,the accuracy reaches 86%,and the recall rate reaches 83%.Compared with the traditional classification model,the results are respectively improved by 5%,7% and 5%.The AUC value of the ROC curve is up to 0.88,and the overall classification performance is improved.The main contributions and innovations of this paper are as follows:(1)This paper proposed a WeChat data analysis model which is based on machine learning and word2 vec.By using the optimization method of machine learning system,the model is trained and optimized,which improves the system classification performance.And the sentiment analysis of WeChat text is realized.(2)The data acquisition function of WeChat platform is designed and implemented,which solves the problem that the traditional web crawler server is limited in frequency by using the method of Python crawler and Hook technology.Through the parallel collection of multiple mobile terminals,the data acquisition rate is improved,and the large-scale acquisition of WeChat data is realized,which provides data support for subsequent experimental study.