Research On Network Public Opinion Classification Algorithm Based On Machine Learning

With the rapid development of Internet social platforms,some extreme remarks,online rumors and negative thoughts often appear on all kinds of news,microblog,wechat and twitter,which has seriously affected social stability and unity.How to mine public opinion information from these media information texts is becoming more and more important for effective monitoring and management of network public opinion.Therefore,this paper proposes an improved public opinion information classification algorithm based on machine learning.The main work is as follows:(1)For the unstructured data of network public opinion,we decided to use the method of automatic machine classification to extract public opinion information keywords from short texts.According to the existing Word embedding principle,the characteristics of network public opinion are encoded by Word embedding,and the one hot and word2 vec methods are used to realize the input of public opinion information context,keyword prediction and Word embedding coding of public opinion information.This paper discusses the essential meaning of public opinion sequence information from local sequence to global sequence,analyzes it combined with word vector and public opinion semantic features,and preprocesses the network public opinion information,so as to construct the public opinion keyword extraction algorithm.(2)Aiming at the problems of sparse information and low accuracy of training text features in the existing models after word segmentation of network public opinion information,a ml-svm support vector machine algorithm based on machine learning is proposed.Through the characteristics of public opinion information and Shannon’s law,the discovery mechanism of public opinion neologisms is established,which solves the problem of insufficient characteristics of training text set;In order to better measure the affinity between public opinion information,a public opinion text information updated algorithm is established,and the effectiveness of this algorithm is verified by simulation experiments and comparative analysis on different data sets.Finally,the classification algorithm proposed in this paper is applied to "social public opinion hot spot analysis",and nine different categories are tested,and the corpus data of different scales are compared and analyzed.The results show that the F1 measurement value of ml-svm classification model reaches 95.2% and 96.6% respectively in terms of recall rate and accuracy rate.It is verified that the network public opinion classification algorithm based on machine learning proposed in this paper has good performance.22 Figures,7 Tables,60 References.