| With the rapid development of Internet,microblog became the most popular social networking platform for its characteristic of fast,short,and casual.A large number of user groups make microblog very far-reaching influence on society.Through this platform,the users are keen on real-time published personal status and experiences,which contains the viewpoint of characters,events,or product evaluation,But great value is hidden in these information which is seemingly trivial and complex,through the analysis,can dig out the user's emotional tendency.This information has a very big help and promote role for both the business marketing and government surveillance of public opinion.Therefore,weibo sentiment analysis research has become a focus of research.But because of the characteristic of the microblog and the explosive growth of the huge amounts of data,the microblogs emotion analysis and research are face with great challenges and tests.In this paper,the Chinese microblogging sentiment analysis was studied,and in reference on the basis of traditional method,and proposes a microblog sentiment analysis algorithm based on Hadoop,and finally implemented the analysis system,ultimately found a distributed and parallel way of massive data processing for microblog sentiment analysis,put forward suggestions and references for related studies.The concrete research content is summarized as: first use the ICTCLAS tool for text participle and feature extraction,and establish 2-POS features of parts of speech and word frequency features;Then add emoticons,negative word library and degree adverb wordlist on the basis of the existing Emotional dictionary,again with the improved weighted voting classifier combination for subjective and objective texts;Finally build Hadoop platform and implementation the algorithm for emotions analysis of positive and negative.Implement parallelization classification,can improves the classification accuracy and efficiency. |
PDF Full Text Request