Global Neural CCG Parsing with Optimality Guarantees

TL;DR

This paper presents a global neural parsing model for CCG that guarantees optimality during decoding by directly searching the entire subtree space, leading to improved accuracy and efficiency.

Contribution

It introduces the first global recursive neural CCG parser with optimality guarantees, combining global features with efficient A* search, and achieves state-of-the-art results.

Findings

Achieves 0.4 F1 improvement over previous best

Finds optimal parse for 99.9% of sentences

Explores only about 190 subtrees on average

Abstract

We introduce the first global recursive neural parsing model with optimality guarantees during decoding. To support global features, we give up dynamic programs and instead search directly in the space of all possible subtrees. Although this space is exponentially large in the sentence length, we show it is possible to learn an efficient A* parser. We augment existing parsing models, which have informative bounds on the outside score, with a global model that has loose bounds but only needs to model non-local phenomena. The global model is trained with a new objective that encourages the parser to explore a tiny fraction of the search space. The approach is applied to CCG parsing, improving state-of-the-art accuracy by 0.4 F1. The parser finds the optimal parse for 99.9% of held-out sentences, exploring on average only 190 subtrees.