Selecting Decomposable Models for Word-Sense Disambiguation: The Grling-Sdm System

This paper describes the grling-sdm system, which is a supervised probabilistic classifier that participated in the 1998 SENSEVAL competition for word-sense disambiguation. This system uses model search to select decomposable probability models describing the dependencies among the feature variables. These types of models have been found to be advantageous in terms of efficiency and representational power. Performance on the SENSEVAL evaluation data is discussed.