Application of Polar Similarity Computing in Words Orientation Identification
In this paper, a new method of Chinese words orientation computing based on HowNet is presented. It uses the subjective sememe of HowNet to calculate the polar similarity between concepts, and uses word co-occurrence window to calculate the polar similarity between candidate word and baseline words, then computes words sensibility degree to identify words sentimental orientation. Experimental results show that the proposed methods can identify words semantic orientation effectively, and it performs best in the subtasks of polarity identification of orientation words in the First Chinese Opinion Analysis Evaluation.
subjective sememe polar similarity polarity
Le Song Tingting He
Department of Computer Science, HuaZhong Normal University, Network Media Branch, National Language Department of Computer Science, HuaZhong Normal University, Network Media Branch. National Language
国际会议
重庆
英文
231-234
2011-01-21(万方平台首次上网日期,不代表论文的发表时间)