Research On Text Sentiment Analysis For Wechat Public Platform

In recent years,text data analysis is popular in major Internet sites,and a lot of branch technologies are focused on network public opinion monitoring and analysis system.As the number of Internet online users has been increasing,the internal capacity of large data is expanding,and then a network public opinion platform is formed with the theme of social networking.The public opinion monitoring based on micro-blog is more extensive,and the technology is more mature.The public opinion based on the WeChat platform is less,and more importantly,the WeChat platform has become a more and more important gathering place of public opinion.This paper is based on the application of the WeChat platform.Through the analysis of the data and the needs of the users,a complete public opinion monitoring system architecture which can be popularized in the WeChat public platform is designed.In this paper,the WeChat data acquisition system is designed and implemented.The information of WeChat public platform is crawled from the Internet,and the related denoising technology can effectively reduce the redundancy of information,improve the authenticity and reliability of the data,and provide data support for the follow-up research.This paper proposes an emotion analysis model,which consists of two parts: one is the word vector Doc2 vec model,the other is the machine learning classification algorithm.The Word2 vec word vector model does not take into account the influence of ordering on semantic analysis,but only semantic analysis of words with fixed dimensions.The Doc2 vec model shares the word vector with the only paragraph vector in the prediction,and infers the paragraph vectors through the fixed word vector and training new paragraph vectors.The model contains more abundant semantic information.The support vector machine(SVM)algorithm uses the inner product kernel to replace the nonlinear mapping to the high dimensional space.The decision function is determined by a few training support vectors.To some extent,it avoids the dimension disaster and helps us grasp the key sample,and has good robustness.The feasibility of combining the vector model with a variety of machine learning algorithms is verified through experiments,and the optimal parameters of the model are obtained through experimental comparison and optimization,and the emotional analysis of the public public opinion of the WeChat public is realized.This design completed a complete set of WeChat platform information monitoring system,including the WeChat platform text online collection system,text emotional analysis system,public opinion display system,the whole system can better meet the needs of public opinion monitoring and analysis.