Verb Class Discovery from Rich Syntactic Data

@incollection{citeulike:2637717, abstract = {Previous research has shown that syntactic features are the most informative features in automatic verb classification. We investigate their optimal characteristics by comparing a range of feature sets extracted from data where the proportion of verbal arguments and adjuncts is controlled. The data are obtained from different versions of valex [1] – a large scf lexicon for English which was acquired automatically from several corpora and the Web. We evaluate the feature sets thoroughly using four supervised classifiers and one unsupervised method. The best performing feature set includes rich syntactic information about both arguments and adjuncts of verbs. When combined with our best performing classifier (a novel Gaussian classifier), it yields the promising accuracy of 64.2\% in classifying 204 verbs to 17 Levin (1993) classes. We discuss the impact of our results on the state-or-art and propose avenues for future work.}, author = {Sun, Lin and Korhonen, Anna and Krymolowski, Yuval }, citeulike-article-id = {2637717}, doi = {http://dx.doi.org/10.1007/978-3-540-78135-6\_2}, journal = {Computational Linguistics and Intelligent Text Processing}, keywords = {classifier, clustering, frames, gaussian, semantic, subcategorization, verb}, pages = {16--27}, posted-at = {2008-04-07 14:17:30}, priority = {0}, title = {Verb Class Discovery from Rich Syntactic Data}, url = {http://dx.doi.org/10.1007/978-3-540-78135-6\_2}, year = {2008} }

TY - CHAP ID - citeulike:2637717 L3 - citeulike-article-id:2637717 TI - Verb Class Discovery from Rich Syntactic Data JF - Computational Linguistics and Intelligent Text Processing SP - 16 EP - 27 N2 - Previous research has shown that syntactic features are the most informative features in automatic verb classification. We investigate their optimal characteristics by comparing a range of feature sets extracted from data where the proportion of verbal arguments and adjuncts is controlled. The data are obtained from different versions of valex [1] – a large scf lexicon for English which was acquired automatically from several corpora and the Web. We evaluate the feature sets thoroughly using four supervised classifiers and one unsupervised method. The best performing feature set includes rich syntactic information about both arguments and adjuncts of verbs. When combined with our best performing classifier (a novel Gaussian classifier), it yields the promising accuracy of 64.2% in classifying 204 verbs to 17 Levin (1993) classes. We discuss the impact of our results on the state-or-art and propose avenues for future work. KW - classifier KW - clustering KW - frames KW - gaussian KW - semantic KW - subcategorization KW - verb AU - Sun, Lin AU - Korhonen, Anna AU - Krymolowski, Yuval PY - 2008/// UR - http://dx.doi.org/10.1007/978-3-540-78135-6_2 ER -