| Paper Details: | Downloads: 720 |
| Serial Number: | P1121132687
|
| Title: | Improving Text Sentiment Classification Performance by Balancing Positive and Negative Features
|
| Authors: | Long-Sheng Chen, Chia-Wei Chang, Chun-Chin Hsu
|
| Abstract: | With the rapid development of new text based media such as blogs, twitters, and etc, more and more people express their opinions and comments about purchased products or services. These online user experiences especially those harmful evaluations of products will spread quickly and indeed influence consumer behaviours of potential customers. Therefore, how to effectively identify users’ sentiments becomes one of critical issues. In recent years, machine learning algorithms have been considered as one of effective solutions for sentiment classification. However, with the increasing amount of the online reviews, the feature space of textual data increases dramatically. The performances of machine learning methods have been degraded due to this dimensionality problem. Consequently, this study aims to propose a novel feature selection strategy called Category Features Oriented (CFO) method which equally selects features from both positive and negative sentiments. Moreover, support vector machines (SVM) have been employed to construct classifiers for identifying bloggers’ sentiments. Finally, one case study from real world blogs will be provided to illustrate the effectiveness of our proposed CFO approach. Compared with traditional methods including Information Gain (IG) and Fisher linear discriminant analysis (FLDA), experimental results indicated that the proposed method can improve sentiment classification performance.
|
| Keywords: | Feature selection, Sentiment Classification, Support vector machines, Text mining.
|
| Journal/Conference: | ICGST Conference on Computer Science and Engineering, CSE-11
|
| Volume: |
|
| Issue: |
|
| Submission Date: | 8/9/2011 12:00:00 AM
|
| Review Date: | 11/8/2011 12:00:00 AM
|
| Publishing Date: | 12/19/2011 12:00:00 AM
|
| Article Downloads: | 720
|
| Download: |
|