Design And Implementation Of Sentiment Analysis System Based On Native Bayes Algorithm

With the rapid development of the Internet,users can freely express and share their feelings and views on the social network platform.The emotional analysis of social network information can excavate the huge commercial and public opinion value behind it.In social networking,micro-blog is different from traditional text sentiment analysis due to its free and complex language format,and blog is similar to traditional text analysis.Based on this,the social network in this paper is mainly micro-blog text.At present,the emotional analysis of social networks is in the initial and developing stages,and there are a lot of problems to be solved.Therefore,the research of emotional analysis based on social network is of theoretical significance and application value.The system of public opinion information of social network analysis,get on a topic on micro-blog micro-blog information as public information through the web crawler,through Chinese segmentation,information extraction,statistical analysis,and finally to the graphics and tables are displayed,so as to provide the basis for decision making.This paper mainly provides a solution to the construction of the emotional analysis system combining the feature extraction of emotional dictionary and the Bias classification algorithm.This paper mainly completed three aspects of the work:1.The construction method of emotion analysis system at home and abroad is analyzed,and the idea of emotional analysis system for constructing social network information is clarified.the detailed analysis of the social network information emotional analysis system is made,and the main frame and various functional modules of the system are given.This paper introduces the technical principle of each module of the system and implementation of sentiment analysis,from two aspects of theory and practice to introduce the application scene,the core technology,including technology(web crawler collection and DOM tree structured data extraction of micro-blog and Natural Language Processing technology using crawler)(mainly used NLPIR)etc..2.This paper analyzed the CHI feature selection algorithm,according to the classification accuracy of feature selection algorithm was proposed based on the algorithm of emotion dictionary,focus on building a comprehensive sentiment dictionary based on SO-PMI algorithm,constructed based emotion dictionary according to different lexical features,emoticons sentiment dictionary and network language emotion dictionary,the three a dictionary is integrated into the final emotion dictionary,the dictionary can be used for feature extraction and analysis of emotion.The emotional dictionary is applied to the participle,and the result of the participle is regarded as an emotional feature item.Three algorithms of word frequency,BOOL and TF-IDF are applied to the calculation of emotional weight.3.The concrete realization process of the emotional analysis system of social network information is explained,and the details of the realization of each function module are explained.Because micro-blog has the characteristics of short text,and uses naive Bayes algorithm to do sentiment analysis,it divides micro-blog text into two parts:the overall view and the split view according to the opinion segmentation.finally,based on the above theory and emotion analysis method,the visualization of the emotional analysis system is realized.