| Summary: | 碩士 === 國立臺北科技大學 === 資訊工程系研究所 === 103 === The research of sentiment analysis aims to explore the sentiment words which are the feelings that the authors were expressing. Several existing researches focus on author’s emotions. News articles are usually described in objective terms. Therefore, it is difficult to find the sentiment words in the articles.
In this paper, we propose to find the reader’s feeling when they read Chinese news. We use feature selection methods, such as document frequency thresholding, mutual information and Chi-square Test, to select the candidate words for sentiment classification. Then, we evaluate the performance of sentiment classification with SVM classifiers. The experimental results show the effectiveness of the proposed feature selection methods in news sentiment classification.
|
|---|
