A Span-based Linearization for Constituent Trees

TL;DR

This paper introduces a new span-based linearization method for constituent trees, combined with a locally normalized model that is fast, interpretable, and achieves competitive accuracy on standard benchmarks.

Contribution

The paper presents a novel span-based linearization approach and a locally normalized model that outperforms existing local models and rivals global models in efficiency and accuracy.

Findings

Achieves 95.8 F1 on PTB

Achieves 92.4 F1 on CTB

Faster and more interpretable than previous models

Abstract

We propose a novel linearization of a constituent tree, together with a new locally normalized model. For each split point in a sentence, our model computes the normalizer on all spans ending with that split point, and then predicts a tree span from them. Compared with global models, our model is fast and parallelizable. Different from previous local models, our linearization method is tied on the spans directly and considers more local features when performing span prediction, which is more interpretable and effective. Experiments on PTB (95.8 F1) and CTB (92.4 F1) show that our model significantly outperforms existing local models and efficiently achieves competitive results with global models.