Measuring Semantic Relatedness between Words Using Lexical Context

Semantic relatedness measurement between words is always a hot issue interested by many researchers. It can be applied to various tasks of NLP and IR with a big challenge. We propose a method for measuring semantic relatedness between words using lexical context in this paper. The method can mainly be divided into two steps. Firstly, for each word of a word-pair, a lexical context is generated exploiting search engine with WordNet, which is constituted by the words highly related to the target word. Secondly, semantic relatedness between words is measured by considering semantic relatedness between a word and lexical context of another word for an original word-pair. Experimental results on Miller-Charles benchmark dataset show our proposed method outperforms all other state of the art related approaches, achieving a Pearson correlation coefficient of 0.912. It shows more competitive than other methods.