Efficiently identifying semantic orientation algorithm for Chinese words
Zhongming Han, Qian Mo, Min Zuo, Dagao Duan · 2010
In this paper, two methods for effective semantic orientation identification algorithm are proposed, which are based on HowNet. Some criteria for computing the orientation similarity between given word and benchmark words set are compared and a k-NN classification for determining semantic orientation for given word is proposed. Comprehensive experiments are conducted and the result show the k-NN method can effectively identify common words semantic orientation and accuracy rate has greatly been improved arriving at 96.5%.