Semantic Word Similarity Learned from Heterogenous Knowledge Bases

Abstract

AbstractRecognition of semantic similarity between words plays an important role in text information management, information retrieval, and natural language processing. There are two major approaches to recognizing the semantic similarity, among which one way is extracting similarity relationships based on a structured semantic dictionary, while the other way is learning the semantic similarity from a large corpus. Building a semantic dictionary is a time-consuming task which also requires much expertise, while the learning method alone cannot extract precise similarity between words. This paper proposes to expand the semantic dictionary by learning the word similarity from heterogenous knowledge bases statistically. This method can not only expand the semantic dictionary from the open knowledge bases but also achieve accurate semantic similarity. In the evaluation of semantic relatedness competition held by CCF, the proposed system ranks the 3rd place according to the macro-average F1 and the 2nd place according to the micro-average F1.KeywordsHeterogeneous Knowledge BasesSimilar WordsWord Semantic SimilarityOpen Knowledge BaseSemantic LexiconThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.