Analyzing Chinese Sentence Patterns for Feature-Based Opinion Mining

• Main Authors: Wen-Chi Cheng, 鄭文綺
• Other Authors: Ching-Sheng Hsu
• Format: Others
• Language: en_US
• Published: 2012
• Online Access: http://ndltd.ncl.edu.tw/handle/95030448759311890386

碩士 === 銘傳大學 === 資訊管理學系碩士班 === 100 === This study discovers part-of-speech (POS) patterns of opinion sentences in Chinese reviews. These patterns are able to identify opinion sentences, feature words, and opinion/feeling words. Additionally, frequently used degree words and negation words were collected. They are useful to determine opinion orientations and strength. In order to identify opinion targets, the associations between opinion/feeling words, feature words, and corresponding features were discovered. An algorithm for feature-based summarization was proposed based on the patterns and association rules. Both car and movie reviews were collected for training and testing and the experimental results demonstrate that the proposed approaches perform well with Chinese product reviews particularly in movie reviews.