Shift-Reduce Constituent Parsing with Neural Lookahead Features

TL;DR

This paper introduces a neural lookahead feature for shift-reduce constituent parsing, significantly improving accuracy by utilizing full sentence information to predict constituent hierarchies.

Contribution

It proposes a bidirectional LSTM-based lookahead feature that enhances transition-based parsing by incorporating non-local constituent information.

Findings

Achieved 1.3% absolute improvement on WSJ dataset

Achieved 2.3% absolute improvement on CTB dataset

Set new state-of-the-art accuracies for fully-supervised parsing

Abstract

Transition-based models can be fast and accurate for constituent parsing. Compared with chart-based models, they leverage richer features by extracting history information from a parser stack, which spans over non-local constituents. On the other hand, during incremental parsing, constituent information on the right hand side of the current word is not utilized, which is a relative weakness of shift-reduce parsing. To address this limitation, we leverage a fast neural model to extract lookahead features. In particular, we build a bidirectional LSTM model, which leverages the full sentence information to predict the hierarchy of constituents that each word starts and ends. The results are then passed to a strong transition-based constituent parser as lookahead features. The resulting parser gives 1.3% absolute improvement in WSJ and 2.3% in CTB compared to the baseline, given the highest reported accuracies for fully-supervised parsing.