Increasing Density through New Relations and PoS Encoding in WordNet.PT

Authors: Raquel, Amaro; Sara, Mendes; Palmira, Marrafa
Year: 2013
Venue: International Journal of Computational Linguistics and Applications, ISSN 0976-0962, vol.4, nr. 1, pp. 11-27.(also in Proc. of the 14th International Conference on Intelligent Text Processing and Computational Linguistics (CICLing 2013), Samos, Greece, March 24-30 2013
Link: http://www.gelbukh.com/ijcla/2013-1/IJCLA-2013-1-pp-011-027-01-Increasing.pdf
Product of the Action: No

Abstract:
This paper reports research developed in the scope of building a wordnet for Portuguese (WordNet.PT), particularly focusing on the impact the results obtained have in the density of the network of relations and, thus, on its usability for NLP tasks. Following from basic research on different linguistic phenomena and on strategies for modeling them in relational models of the lexicon, the implementation of these results amounts to a richer resource, with new cross-PoS relations and information on event and argument structures, thus crucially contributing to accurately modeling all the main PoS in the database. We also define a way to integrate prepositions in wordnets and discuss the motivations and modeling strategies used to do so. Based on this work, we show how our contributions augment the coverage and the accuracy of WordNet.PT, by increasing the density of the network of relations, thus making it more usable for NLP applications.