knowledge acquisition. Ide et al. (2001 and 2002), Ng et al. (2003), and Diab (2003, 2004a, and 2004b) made research on the use of align-ment for WSD. Diab and Resnik (2002) investigated the feasi-bility of automatically annotating large amounts of data in parallel corpora using an unsupervised algo[r]
operate on the existing rules of the grammar, with-out introducing competing rules. These features aretreated, both in feature-weight tuning and in decod-ing, on the same footing as the rest of the model,allowing it to weigh the WSD model predictionsagainst other pieces of evidence so as to optimize[r]
one and only one sense to a word in context (and hence the produce the test material itself) than to perform other NLP related tasks. One of the present authors has discussed Kilgarriff's figures elsewhere (Wilks, 1997) and argued that they are not, in fact, as gloomy as he suggests. A[r]
suspect that selecting the appropriate levelof abstraction could be on between bothlevels. We use a very simple method forderiving a small set of appropriate mean-ings using basic structural properties ofWordNet. We also empirically demon-strate that this automatically derived set ofmeanings groups[r]
abilistic classifiers for word-sense disambiguation that offers advantages over previous approaches. Most pre- vious efforts have not attempted to systematically iden- tify the interdependencies among contextual features (such as collocations) that can be used to classify the me[r]
James Cussens, David Page, Stephen Muggleton, and Ashwin Srinivasan. 1997. Using Inductive Logic Programming for Natural Language Processing. Workshop Notes on Empirical Learning of Natural Language Tasks, Prague, pages 25-34. Hoa T. Dang and Martha Palmer. 2005. The Role of Semantic Roles in Disamb[r]
which is most related to the context words.Relatedness between word senses is mea-sured using the WordNet::Similarity Perlmodules.1 IntroductionMany words have different meanings when used indifferent contexts. Word Sense Disambiguation isthe task of identifying the inten[r]
however, as such isolated tokens tend to strongly fa- vor a particular sense (the less "bursty" one). We have yet to use this additional information. 8 Evaluation The words used in this evaluation were randomly selected from those previously studied in the litera- ture. They include words whe[r]
Santa Cruz, California, USA.Franz Josef Och and Hermann Ney. 2000. Improvedstatistical alignment models. In Proceedings of the38th Annual Meeting of the Association for Com-putational Linguistics (ACL), pages 440–447, HongKong.Martha Palmer, Christiane Fellbaum, Scott Cotton,Lauren Delfs, and Hoa Tr[r]
sense-disambiguation questions. FUTURE WORK Although our results are promising, this particular method of assigning senses to words is quite limited. It assigns at most two senses to a word, and thus can extract no more than one bit of information about the translation of that <[r]
sense disambiguation either embed topic featuresin a supervised model (Cai et al., 2007) or relyheavily on the structure of hierarchical lexiconssuch as WordNet (Boyd-Graber et al., 2007). Inthis paper, we propose a novel framework whichis fairly resource-poor in that it requires only[r]
WASPBENCH: a lexicographer's workbench supporting state-of-the-artword sense disambiguation.Adam Kilgarriff, Roger Evans, Rob KoelingMichael Runde11, David TugwellITRI, University of BrightonFirstname.Lastname@itri.brighton.ac.uk1 BackgroundHuman Language Technologies (HLT) need dic-ti[r]
work on word-sense disambiguation, which extends the approach of using massive lexicographic resources (e.g., parallel corpora, dictionaries, thesauruses and encyclopedia) in order to attack the knowledge- acquisition bottleneck that Bar-Hillel identified over thirty years ago.[r]
word senses are derived automatically fromword alignments on a parallel corpus. We builtfive classifiers with English as an input lan-guage and translations in the five supportedlanguages (viz. French, Dutch, Italian, Span-ish and German) as classification output. Thefeature vectors incorporate b[r]
frequent ones. However, the difference betweenour complex model and the best HMM (EM withsmoothing and random restarts, 55%) is not signifi-cant.The best (supervised) system in the SemEval task(Ye and Baldwin, 2007) reached 69% accuracy. Thebest current supervised system we are aware of(Hovy et al.,[r]
ing, where each sense tagged occurrence of a particu-lar word is transformed into a feature vector, which isthen used in an automatic learning process. The appli-cability of such supervised algorithms is however lim-ited only to those few words for which sense taggeddata is avai[r]
Proceedings of EACL '99 Word Sense Disambiguation in Untagged Text based on Term Weight Learning Fumiyo Fukumoto and Yoshimi Suzukit Department of Computer Science and Media Engineering, Yamanashi University 4-3-11 Takeda, Kofu 400-8511 Japan {fukumoto@skye.esb, ysuzuki@winderme[r]
A Kernel PCA Method for Superior Word Sense DisambiguationDekai WU1Weifeng SU Marine CARPUATdekai@cs.ust.hk weifeng@cs.ust.hk marine@cs.ust.hkHuman Language Technology CenterHKUSTDepartment of Computer ScienceUniversity of Science and TechnologyClear Water Bay, Hong KongAbstractWe intr[r]
Our approach of exploiting parallel texts for word sense disambiguation consists of four steps: (1) parallel text alignment (2) manual selection of tar-get translations (3) training of WSD classifier (4) WSD of words in new contexts. Parallel Text Alignment In this step, paralle[r]
accuracy. In this paper, we show that using wellcalibrated probabilities to estimate sense priors isimportant. By calibrating the probabilities of thenaive Bayes algorithm, and using the probabilitiesgiven by logistic regression (which is already wellcalibrated), we achieved significant improv[r]