Non-Projective Dependency Parsing with Non-Local Transitions

TL;DR

This paper introduces a novel non-projective dependency parser with non-local transitions that directly create long-distance arcs, reducing error propagation and achieving state-of-the-art accuracy among greedy algorithms.

Contribution

It proposes a new transition system based on the Covington parser that improves efficiency and accuracy in non-projective dependency parsing.

Findings

Outperforms the original Covington parser

Achieves best accuracy on Stanford Dependencies conversion

Reduces error propagation in dependency parsing

Abstract

We present a novel transition system, based on the Covington non-projective parser, introducing non-local transitions that can directly create arcs involving nodes to the left of the current focus positions. This avoids the need for long sequences of No-Arc transitions to create long-distance arcs, thus alleviating error propagation. The resulting parser outperforms the original version and achieves the best accuracy on the Stanford Dependencies conversion of the Penn Treebank among greedy transition-based algorithms.