Evaluation of Dependency Parsers on Unbounded Dependencies

We evaluate two dependency parsers, MSTParser and MaltParser, with respect to their capacity to recover unbounded dependencies in English, a type of evaluation that has been applied to grammarbased parsers and statistical phrase structure parsers but not to dependency parsers. The evaluation shows that when combined with simple post-processing heuristics, the parsers correctly recall unbounded dependencies roughly 50% of the time, which is only slightly worse than two grammar-based parsers specifically designed to cope with such dependencies.

• Type: Journal
• Year: 2010
• Where: COLING
• Authors: Joakim Nivre, Laura Rimell, Ryan T. McDonald, Carlos Gómez-Rodríguez
• Keywords: COLING 2010, Computational Linguistics, Dependency Parsers, Parsers, Unbounded Dependencies