UMCC_DLSI: Reinforcing a Ranking Algorithm with Sense Frequencies and Multidimensional Semantic Resources to solve Multilingual Word Sense Disambiguation

Second Joint Conference on Lexical and Computational Semantics (*SEM), Volume 2: Seventh International Workshop on Semantic Evaluation (SemEval 2013), pages 241–249, Atlanta, Georgia, June 14-15, 2013. c⃝2013 Association for Computational Linguistics UMCC_DLSI: Reinforcing a Ranking Algorithm with Sense Frequencies and Multidimensional Semantic Resources to solve Multilingual Word Sense Disambiguation Yoan Gutiérrez, Yenier Castañeda, Andy González, Rainel Estrada, Dennys D. Piug, Jose I. Abreu, Roger Pérez Antonio Fernández Orquín, Andrés Montoyo, Rafael Muñoz Franc Camara DI, University of Matanzas DLSI, University of Alicante Independent Consultant Matanzas, Cuba Alicante, Spain USA {yoan.gutierrez, yenier.castaneda, rainel.estrada, dennys.puig, jose.abreu, roger.perez}@umcc.cu, andy.gonzalez@infonet.umcc .cu antonybr@yahoo.com, {montoyo,rafael}@dlsi.ua. es info@franccamara.c om Abstract This work introduces a new unsupervised approach to multilingual word sense disambiguation. Its main purpose is to automatically choose the intended sense (meaning) of a word in a particular context for different languages. It does so by selecting the correct Babel synset for the word and the various Wiki Page titles that mention the word. BabelNet contains all the output information that our system needs, in its Babel synset. Through Babel synset, we find all the possible Synsets for the word in WordNet. Using these Synsets, we apply the disambiguation method Ppr+Freq to find what we need. To facilitate the work with WordNet, we use the ISR-WN which offers the integration of different resources to WordNet. Our system, recognized as the best in the competition, obtains results around 69% of Recall. 1 Introduction Word Sense Disambiguation (WSD) focuses on resolving the semantic ambiguity of a given word. This is an important task in Natural Language Processing (NLP) because in many applications, such as Automatic Translation, it is essential to know the exact meaning of a word in a given context. In order to solve semantic ambiguity, different systems have been developed. However, we can categorize them in two main groups: supervised and unsupervised systems. The supervised ones need large quantity of hand-tagged data in order to gather enough information to build rules, train systems, and so on. Unsupervised systems, on the other hand, do not need such a large amount of hand-tagged datasets. This means that, when there aren’t enough corpora to train the systems, an unsupervised system is a good option. A sub-task of WSD is Multilingual Word Sense Disambiguation (MWSD) (Navigli et al., 2013) that aims at resolving ambiguities in different languages. In a language, there are words that have only one sense (or meaning), but in other languages, the same words can have different senses. For example, “patient” is a word that in English can be either a noun or an adjective, but in German, it only has one sense - “viz” (a person that needs treatment). This shows that the information obtained by combining two languages can be more useful for WSD because the word senses in each language can complement each other. For it to be useful, MWSD needs a multilingual resource that contains different languages, such as BabelNet (Navigli and Ponzetto, 2010; 2012) and EuroWordNet (Vossen, 1998). 241 As the preferred …

Content truncated to 500 words