Measurement of Semantic Similarity Between Words: A Survey

Abstract

Semantic similarity measures between words play an important role in community mining, document clustering, information retrieval and automatic metadata extraction. For a computer to decide the semantic similarity between words, it should understand the semantics of the given words. Computer is a syntactic machine, which cannot understand the semantics. So it always made an attempt to represent the semantics words as syntactic words. Today, there are various methods proposed for finding the semantic similarity between words. Some of these methods have used the information sources as precompiled databases like WordNet and Brown Corpus. Some are based on Web Search Engine. In this paper we have described the methods based on precompiled databases like WordNet and Brown Corpus as well as the web search engine. Along with this we have compared the all methods on the basis of performance and their limitation. From the study, Experimental result on Miller-Charles benchmark dataset show that the method by the Danushka Bollegala, Yutaka Matsuo, and Mitsuru Ishizuka based on web search engine results outperforms all the existing semantic similarity measures by a wide margin, achieving a correlation coefficient of 0.87