The Design And Realization Of The Sentiment Analysis System Based On The Dependency Grammar

With the rapid development of Internet technology,as a carrier of information,the Internet has become the main channel to obtain information and express own subjective views of countless users.The technology which combines computer technology and natural language processing technology has become the main requirements of some comercial organizations and related departments of goverment,which is used to recognize and analyze the viewpoint of Internet users.Therefore,this is why the sentiment analysis system came out.Combined with the actual requirements,the the sentiment analysis system base on the dependency grammar purposed in this paper mainly analyse sentiment tendency of the comments online.The data which is need by the sentiment analysis system have through web crawl,text extraction,Chinese word segmentation and part-of-speech tagging.With the text data,the sentiment analysis system detect the topic first,and then to have the hot and sensitive topic identification.through the above this paper improve the efficiency of the subsequent text sentiment analysis.Before the sentiment analysis,this paper use the network phrase dictionary and Common Wrong character dictionary to expand the sentiment dictionary which is baesd on the NTUSD and HowNet.The method which is taken by this paper is combine the knowledge of natural language processing,including dependency grammar analysis,Semantic role labeling,and Named entity recognition,to extra the pair of word which are feature word and sentiment word,and then extract the list of modifier word.After that,calculating the polarity of sentiment to complete the sentiment analysis of the network comment.In this paper,the reslut of a experiment which is based on the real network dataset have proved this method which is based on the dependency grammar to analysis the sentiment tendency is feasible.