Text sentiment analysis based on multi-granularity joint solution

With the increase of the number of comments on major websites, how to distinguish the emotional tendencies of these comments has become the focus of researches. Through the analyzing and studying of these comments, a sentiment analysis method based on multi-granularity joint solution is proposed in this paper. In order to find the accurate evaluating objects, obtaining the emotional vocabularies and mining the relative evaluating objects hiding behind these vocabularies have been done at first; then a multi-granularity joint solution model is built; Furthermore, the scores of each comments are given by this model. Finally, some experiments to test the validity and accuracy of our proposed method have been executed, the results show that the average accuracy rate has been improved by 4% compared with the traditional multi - granularity methods.