Poly-co: a multilayer perceptron approach for coreference detection

Abstract

This paper presents the coreference resolution system Poly-co submitted to the closed track of the CoNLL-2011 Shared Task. Our system integrates a multilayer perceptron classifier in a pipeline approach. We describe the heuristic used to select the pairs of coreference candidates that are feeded to the network for training, and our feature selection method. The features used in our approach are based on similarity and identity measures, filtering in-formations, like gender and number, and other syntactic information.