Paper: Improving Graph-based Dependency Parsing with Decision History

ACL ID C10-2015
Title Improving Graph-based Dependency Parsing with Decision History
Venue International Conference on Computational Linguistics
Session Poster Session
Year 2010
Authors

This paper proposes an approach to im- prove graph-based dependency parsing by using decision history. We introduce a mechanismthatconsidersshortdependen- cies computed in the earlier stages of pars- ing to improve the accuracy of long de- pendencies in the later stages. This re- lies on the fact that short dependencies are generally more accurate than long depen- dencies in graph-based models and may be used as features to help parse long de- pendencies. The mechanism can easily be implemented by modifying a graph- based parsing model and introducing a set of new features. The experimental results show that our system achieves state-of- the-art accuracy on the standard PTB test set for English and the standard Penn Chi- nese Treebank (CTB) test set for Chinese.