Defining classifier regions for WSD ensembles using word space features

Abstract

Based on recent evaluation of word sense disambiguation (WSD) systems [10], disambiguation methods have reached a standstill. In [10] we showed that it is possible to predict the best system for target word using word features and that using this 'optimal ensembling method' more accurate WSD ensembles can be built (3-5% over Senseval state of the art systems with the same amount of possible potential remaining). In the interest of developing if more accurate ensembles, we here define the strong regions for three popular and effective classifiers used for WSD task (Naive Bayes - NB, Support Vector Machine - SVM, Decision Rules - D) using word features (word grain, amount of positive and negative training examples, dominant sense ratio). We also discuss the effect of remaining factors (feature-based). © Springer-Verlag Berlin Heidelberg 2006.