Experimental Evaluation

Abstract

Using a corpus-based and computational approach in Chap. 4, I predicted that different clusters can represent different senses via the character similarity clustering analysis and the concept similarity clustering analysis in the automatically computational programming, and I examined the accuracy rates of four target words by my own intuition. In Chap. 5, I then manually evaluated the four target words via sense divisions in Chinese Wordnet and in Xiandai Hanyu Cidian. With both, I was able to obtain higher accuracy rates and higher recalls. In other words, I was able to employ automatically computational programming to predict different senses for chi “eat”, wan2 “play”, huan4 “change”, and shao1 “burn” and examined the accuracy rates by my own intuition. I will next demonstrate that I can use off-line tasks to test my native speakers’ intuitions to support the notion that different clusters divided from a corpus-based and computational approach represent different senses. Different collocation words will affect the interpretations of the four target word. If I can demonstrate that there are several clusters of the related collocation words for chi “eat”, wan2 “play”, huan4 “change”, and shao1 “burn” via off-line tasks, I can predict several different senses for these four target words.