Non-Projective Dependency Parsing with Non-Local Transitions
This work addresses a bottleneck in dependency parsing for natural language processing, though it is incremental as it builds on the Covington parser.
The paper tackles the problem of error propagation in non-projective dependency parsing by introducing a novel transition system with non-local transitions that directly create long-distance arcs, outperforming the original version and achieving the best accuracy on the Stanford Dependencies conversion of the Penn Treebank among greedy transition-based algorithms.
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.