원문정보
초록
영어
Increasing number of subjective text appears on the internet which contains a lot of information. In this paper, we study how to apply supervised learning techniques to solve sentiment classification problems. Using the Tibetan news as data, we find that standard supervised learning techniques definitively outperform human-produced baselines. Moreover, we find that selecting the words with words with polarity as feature, the special syntactic structure such as exclamation sentence pattern, etc. as feature can improve the performance of sentiment classification. Conclusively, the research of sentiment analysis is a more challenging problem.
목차
Abstract
1. Introduction
2. Related Work
2.1. The Supervised Learning Method
2.2. The Unsupervised Learning Method
3. Classification Algorithm
3.1. Naïve Bayes
3.2. Maximum Entropy
4. The Method of Generating Tibetan Text Feature
4.1. The Classification Model
4.3. The Experimental Data
4.4. The Experiments
5. Experimental Results and Analysis
Acknowledgement
References
1. Introduction
2. Related Work
2.1. The Supervised Learning Method
2.2. The Unsupervised Learning Method
3. Classification Algorithm
3.1. Naïve Bayes
3.2. Maximum Entropy
4. The Method of Generating Tibetan Text Feature
4.1. The Classification Model
4.3. The Experimental Data
4.4. The Experiments
5. Experimental Results and Analysis
Acknowledgement
References