Title: A new approach for unsupervised word sense disambiguation in Hindi language using graph connectivity measures

Authors: Amita Jain; D.K. Lobiyal

Addresses: Department of CSE, Ambedkar Institute of Advanced Communication, Technology and Research, Delhi, India ' School of Computer and Systems Sciences, Jawaharlal Nehru University, New Delhi, India

Abstract: Word sense disambiguation (WSD) is an important task in computational linguistics as it is essential for many language understanding applications. In this paper, we propose a graph-based unsupervised WSD method for Hindi text which disambiguates multiple ambiguous words present in the sentence simultaneously. In our approach, we first construct the semantic graph for each interpretation of the given sentence by establishing semantic relations between the pair of words present in the sentence. We use Hindi WordNet to establish semantic relations between the pair of words and then we construct the graph. We find the cost of spanning tree corresponding to each semantic graph and the interpretation for which spanning tree has the minimum cost is identified. This interpretation is considered as the resulting interpretation. Our approach also considers all open class words unlike the previous approaches which focus only on noun.

Keywords: natural language processing; NLP; word sense disambiguation; human computer interface; HCI; Hindi WordNet; graph connectivity; Hindi language; unsupervised WSD; semantic graph; semantic relations.

DOI: 10.1504/IJAISC.2014.065800

International Journal of Artificial Intelligence and Soft Computing, 2014 Vol.4 No.4, pp.318 - 334

Accepted: 12 May 2014
Published online: 29 Nov 2014 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article