Paper: Transition-based Dependency Parsing with Rich Non-local Features

ACL ID P11-2033
Title Transition-based Dependency Parsing with Rich Non-local Features
Venue Annual Meeting of the Association of Computational Linguistics
Session Main Conference
Year 2011
Authors

Transition-based dependency parsers gener- ally use heuristic decoding algorithms but can accommodate arbitrarily rich feature represen- tations. In this paper, we show that we can im- prove the accuracy of such parsers by consid- ering even richer feature sets than those em- ployed in previous systems. In the standard Penn Treebank setup, our novel features im- prove attachment score form 91.4% to 92.9%, giving the best results so far for transition- based parsing and rivaling the best results overall. For the Chinese Treebank, they give a signficant improvement of the state of the art. An open source release of our parser is freely available.