Characterizing the Errors of Data-Driven Dependency Parsing Models

We present a comparative error analysis of the two dominant approaches in datadriven dependency parsing: global, exhaustive, graph-based models, and local, greedy, transition-based models. We show that, in spite of similar performance overall, the two models produce different types of errors, in a way that can be explained by theoretical properties of the two models. This analysis leads to new directions for parser development.

• Type: Conference
• Year: 2007
• Where: EMNLP
• Authors: Ryan T. McDonald, Joakim Nivre
• Keywords: Comparative Error Analysis, EMNLP 2007, Graph-based Models, Natural Language Processing, Transition-based Models