Research On The Sentiment Analysis Method Of Chinese Movie Criticism Based On Internet

With the development of the Internet,more and more users tend to post their own comments on various websites to share their own experiences.These commentary texts express the opinions and emotions of commentators.People are accustomed to obtaining information from various comment texts and looking for some reference opinions.Analysis of the text on the Internet to find out the potential of the information,with great value.However,with the growth of Internet information and the increasing amount of commentary information,people urgently need an automated tool to quickly obtain the information they really need in a large amount of commentary.How to classify the Internet comment texts containing a large amount of information into emotion information and mine the emotion information contained therein are the research difficulties in the field of natural language processing.The related technologies of text emotion analysis came into being.In this paper,the author chooses to use Chinese movie reviews as a research method.Firstly,the comment texts are preprocessed using the Viterbi algorithm based on the hidden Markov model.Subjective and objective texts are classified according to the mutual information between points.),The subjective sentences extracted as the subjective representative of each comment.Then,based on the concept network and the genetic algorithm,this paper proposes a sentiment analysis method that can effectively analyze subjective commentary texts.The conceptual network is used to find the eigenvalues ??of the subjective emotions,and a genetic algorithm is used to train an optimal weight table that transforms disagreements.Based on the eigenvalues ??of emotions and the weight table for converting different opinions,the value of the opinions of each sentence is obtained,and the emotional tendency of each sentence is judged.This paper uses the Chinese movie review information captured by Python to verify subjective and objective text classification methods based on mutual information retrieval between points and sentiment analysis methods based on concept network and genetic algorithm respectively.Experiments prove that the sentiment analysis of the subjective texts in the comment texts and the direct analysis of the comment texts can more accurately find the commentators' opinions.In addition,this paper also presents a method of establishing emotional indicators,which can provide an intuitive and convenient value of real-time emotional indicators for the user reference.