Research On Micro-blog Sentiment Classification Based On Extended Dictionary And Rules

With the rapid development of Internet technology and social network services,the number of micro-blog users in China is increasing,especially Sina micro-blog,and the number of micro-blog texts generated has also increased rapidly.With the introduction and rise of web2.0,the Internet has given netizens more initiative,and micro-blog has penetrated the lives of the vast number of netizens with its simple and flexible features,and has become a platform for users to obtain,share,and publish information,which makes the number of texts containing emotions on the Internet rapid.increase.These texts have great research value in public opinion control,market prediction,etc.,so mining massive micro-blog emotional information technology came into being.Accurately recognizing the sentiment contained in the text of micro-blog and implementing sentiment classification for micro-blog are of great significance both in theoretical research and application.At present,this paper research on dictionary expansion,there is a problem of imperfect sentiment dictionary and low sentiment classification accuracy in sentiment classification research.In order to solve this problem,based on semantic similarity,this paper proposes a new extended dictionary construction method for micro-blog sentiment classification and summarizes sentiment Six combinations of sentiment units are expressed,and then the dictionary is combined with semantic rules to complete the sentiment classification task,effectively identifying the sentiments contained in the microblog text.The main research work includes:First,this paper organize the existing sentiment dictionary resources and propose a method for building a special dictionary for sentiment classification on micro-blog.Combine the two-category dictionary and the multi-category dictionary according to the similarity of words,and organize the emotional dictionary according to the sevencategory sentiment classification system;collect network vocabulary and emoticons,use PMI to determine sentiment to construct the network word dictionary and emoticon dictionary,and explain each part in detail Dictionary construction rules.Then,research on sentiment classification methods,when considering the expansion of sentiment dictionary,the impact of sentiment word scale and sentiment word recognition accuracy on the sentiment classification of micro-blog text is compared,and the influence of similarity threshold on the effectiveness and accuracy of sentiment recognition during dictionary merge is determined to determine the appropriate similarity.The range of degrees.Finally,based on the characteristics of microblog text expression,the shortcomings of sentiment classification method based on sentiment dictionary are analyzed,the combined form of sentiment unit of microblog text is summarized,and a sentiment classification method combining dictionary and rule is proposed.Combining the extended dictionary with text semantic rules,the text sentiment category is obtained by calculating the sentiment intensity value of the sentiment unit.Aiming at the two research topics of expanding dictionaries and combining dictionaries and rules for sentiment classification,corresponding contrast experiments are designed.Experiments prove that the method proposed in this paper has good performance in manual acquisition and public evaluation of data sets.