Automatic semantic role labelling using a memory-based learning system

In this paper we present a semantic role labelling system. The main component of the system is a memory-based classifier. The system has been trained with the Cast3LB-CoNLL-SemRol. The features encode information from dependency syntax. The results (F1 0.86) are comparable with state-of-the-art results (F1 around 0.86) from systems that use information from constituent syntax.