Description and results of the SuperSense tagging task

Abstract

SuperSense tagging (SST) is a Natural Language Processing task that consists in annotating each significant entity in a text, like nouns, verbs, adjectives and adverbs, according to a general semantic taxonomy defined by the WordNet lexicographer classes (called SuperSenses). SST can be considered as a task half-way between Named-Entity Recognition (NER) and Word Sense Disambiguation (WSD): it is an extension of NER, since it uses a larger set of semantic categories, and it is an easier and more practical task with respect to WSD, that deals with very specific senses. We will report on the organization and results of the Evalita 2011 SuperSense Tagging task. © Springer-Verlag Berlin Heidelberg 2013.