Research on sentiment analysis of microblogging based on LSA and TF-IDF
Yingying Li, Bo Shen · 2017
As a typical social network application, the impact of microblogging on people has penetrated into all aspects, which attracts more and more scholars to carry out in-depth study on microblogging. The sentiment analysis of microblogging text is the hot studying field now. Feature selection and extraction is one of the core parts of microblogging text sentiment analysis, TF-IDF algorithm is the most widely used method in selecting features. Although the TF-IDF algorithm is simple to use, there is still a problem of semantic deletion on it, that is, it ignores the semantic information contained in the text. To solve the problem, LSA is introduced in this paper. Firstly, the eigenvectors generated by TF-IDF algorithm is decomposed by singular value. Then, calculating the cosine value between the row vectors of the decomposition results to identify the similarity between the words, which realizes the feature extraction and makes up for the deficiency of TF-IDF. Finally, the extracted features are applied into four classification algorithms to verify the effectiveness of the proposed method. The experimental results show that the introcuction of LSA can make improvements of microblogging text classification in accuracy, recall and F value.