Semantic similarity computation among hindi words using hindi lexical ontology

Abstract

The measurement of semantic similarity between two concepts/words has always been a challenge in the field of document retrieval. As per current literature, there is no method to compute semantic similarity for Hindi words. Therefore, a new computational method is proposed to measure semantic similarity between Hindi words using lexical ontology. In this approach, Hindi WordNet is used which gives the relational information among various Hindi words/concepts. The performance is tested on translated Miller and Charles’s benchmark dataset. In this paper, the three different semantic similarity measuring methods are used to compute semantic similarity for 20 words pairs. The accuracy of the results is also measured in this work using the correlation coefficient. The proposed method in this paper is focusing on the study and analysis of the results obtained from semantic similarity measuring methods on Hindi concepts/words.