Analysis Of Micro-blog Sentiment Orientation Based On SVM

With the advent of the era of Web2.0, web development has entered all areas of people’s lives,in recent years, the emergence of microblogging, make life richer. Microblogging influence growth,attracting a large number of scholars in-depth study of the micro-blog, and emotional wordrecognition and sentiment analysis has become an important issue. In the open microbloggingplatform, providing functions can access information, you can also publish information to otherspoint of view. At the same time, with the release of a wide range of information, along withcreating new problems, such as, the emergence of new words emotion and emotional polarityanalysis of microblogging sentences, new words appear to have a lot of Chinese word difficult toidentify." loose strings "and" debris "; discriminating emotional tendencies when micro blog textsentiment analysis, judgment belongs to positive, negative, neutral judges. Emotional tendency ofthese texts, users can have a grasp of emotions, not only have some commercial value, but alsobeneficial to the community, can also help us to improve in the areas of public opinion monitoring,thesaurus updates, natural language processing.Every day tens of thousands of Chinese microblogging users to refresh the information on themicroblogging emotional word generation and analysis of polar consequent problems, do a goodjob of understanding the user’s attitude is very important and urgent. In this paper, data from theexperiments provided by the airport emotional condition with word recognition, carried tagging,combined with contextual information feature, the feature vector to build on the training corpusdata model construction and testing, and finally get emotional word correct rate (Precision), recall(Recall) and F-values. Emotional words on Weibo effectively correct text recognition isdiscriminating microblogging emotional tendentious premise and foundation. Firstly, combinedwith Chinese information processing, natural language knowledge, combined with laboratoryemotional discovery and analysis of new words emotional tendencies, emotional tendenciesdiscussed various relationships relevant to the analysis done by the establishment of an existing tendency to judge emotion in microblogging the foundation. The ultimate goal of this researchpaper is to improve the accuracy of the data results, recall and F-value, lay the foundation forfurther study.Optional experimental data given microblogging corpus, training and test data to identifyemotional words and emotional orientation analysis is different, experimental results also validatethe methods used in the paper is feasible. Experimental results show that: the method in theemotional word recognition accuracy was34.21%, the recall rate was0.11%, F value of0.002%;results overall recognition rate is not high, but also to lay the foundation for future work. Emotionalsentence polarity discrimination correct rate84.87%recall rate of65.18%, F value of77.27%, thestudy of emotional orientation analysis Chinese microblogging conducted a preliminaryexploration.