SENSEVAL, a computer-based approximation to meaning

Abstract

In this paper we present an overview of the international exercise SENSEVAL, the basic purpose of which is to organize and run evaluation of Word Sense Disambiguation (WSD) algorithms and systems with respect to different words and languages. In order to do this, we first introduce the main goals and the organization of the SENSEVAL exercise; we follow with the methodology developed, the tasks and the linguistic resources created (corpora and dictionaries); and we finally end with some general reflections about the effects on Linguistics and, concretely, on Lexicography motivated by this methodology. We will base on the resources developed for the Spanish SENSEVAL-2 in order to illustrate this methodology.

Keywords

• SENSEVAL
• desambiguación semántica automática
• lingüística computacional