Classification And Mining Of Emotional Information For Online Reviews

With the rapid development of Internet,social network is coming into a new era.Especially the development of media nowadays,information communication is highly participatory and interactive.Since the number of users in the media increase sharply,more and more users on the platform express his own views,the number of online reviews also shows explosive growth.These comments reflect the user's subjective and objective evaluation of an event,including the users' own emotional attitude.Therefore,it is of great significance to analyze the online comment data and to excavate the users' emotional information on the event.Sentiment classification is based on the polarity of user opinion reflected in the text content.It categorizes the text with the same specific emotional tendency into a class,and its purpose is to analyze the theme or other information better that the user reviews pay attention to under the specific emotional orientation.At present,online reviews emotional classification mainly adopts two ways,namely,statistical natural language processing and emotional semantic features.However,there is less research on the combination of these two ways for sentiment classification.Therefore,this paper puts forward the combination of these two ways,and designs an emotion classification method based on the fusion of word vector and emotional ontology.The experiment shows that the classification method proposed in this paper is better than the classification method only based on word vector and the emotion ontology.After sentiment classification of reviews,in order to catch the connection between topic and user sentiment,we propose an emotion information mining model combining Latent Dirichlet Allocation(LDA)and semantic web.The results of the LDA theme analysis are dealt with,and the high frequency words related to the subject are divided into the subject related nouns and emotional words.After that,the word co-occurrence matrix is established by combining thematic words and emotional words,and the semantic web diagram of the comment text is formed based on theword co-occurrence matrix,and the emotional information of the text is described.The experiment show that the model is better than an emotion information mining model based on Term Frequency-Inverse Document Frequency(TF-IDF).