Microblog Emotional Dictionary Built And Application On Sentiment Analysis Of Microblog

In recent years, as an important part of natural language process, sentimentanalysis has attracted many scholars’ attentions. The sentiment analysis of microbloghas become a hot topic studying currently. As a new way of interaction, microbloggives more new charm to the interpersonal communication, and be highly thought ofby the critics. Messages on the micro blog seemingly haphazard, but they have a veryimportant value in the same time. Microblog provides a platform for the users. Userscan reflect their variety social problems by this platform, and release a lot ofinformation with a strong personal emotion or subjective feelings to express their truefeelings.This paper outlines the current situation about the sentiment analysis of text,introduces the various emotional classification models briefly, and summarizes thetraditional sentiment analysis and the characteristics of this new text. Buildingmicroblog emotional dictionary and associated resources is an important work inmicroblog sentiment analysis. In building emotional dictionary, on the one hand, thispaper filters several authoritative emotional dictionaries to get the basic emotionaldictionary; on the other hand, according to the sentence structures of emotional words,this paper builds templates and uses the templates to expand the emotional words.Degree adverbs, conjunctions and negative words have an obvious impact on theemotional words, so the corresponding dictionaries are built in this paper. Emoticonsare commonly used to articulate the current emotion in the microblog. For that reason,this paper extracts the emoticons to build emoticons emotional dictionary. Thenetwork languages which have emotional colors are also extracted to build thenetwork language emotional dictionary. At the same time, some rules which are usedto solve the problems of emotional ambiguity emotional words and implicit emotionalsentences are built. Integrate basic emotional dictionary, extension emotionaldictionary, emoticons emotional dictionary and network language emotional dictionary to form the final microblog emotional dictionary.This paper uses the final microblog emotional dictionary to analysis thesentiment of microblog text. In order to test the microblog emotional dictionary andrules are effective for microblog sentiment analysis, the ME and SVM classificationmodel are selected as two comparison methods. In order to verify the applicability ofthe dictionary, two different kinds of experimental corpus are selected. The balancedexperimental corpus means the sentiment is uniform distributed, but the unbalancedone is randomly distributed. The experimental comparison results show that using themicroblog emotional dictionary and rules is better than the other two classifier modelsboth in the two kinds of experimental corpus. Furthermore, the validity andapplicability of the microblog emotional dictionary, which is built to analysis thesentiment of microblog in this paper, is verified by the experimental results.