Using relative weights to build vector model to classify customer reviews

• Main Authors: Shao-Fan Xie, 解少帆
• Other Authors: Shin-Chieh Chou
• Format: Others
• Language: zh-TW
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/08599593561917008795

碩士 === 國立中央大學 === 資訊管理研究所 === 98 === As the Internet and web2.0 are becoming more and more popular, the number of customer review grows rapidly in the web sites and the blogs. These reviews have gradually become an important reference for the consumers. Therefore, accurate and efficient classification for these reviews will help the consumers to make decisions quickly and correctly. Past studies usually applied the opinion mining or the semantic tendency to classify the reviews. But the significance of the opinion in the diffident types of the reviews may not be the same, and the degree of the semantic tendency may be different. In order to improve the accuracy of classification, this study propose a method that applies the classified documents to get the feedback information, and then calculates the relative weight of the opinion to establish the vector model. The experimental results show that the classification accuracy of relative weight could perform better than the accuracy of semantic weights.