A Structured Span Selector

TL;DR

This paper introduces a grammar-based structured span selector that leverages partial annotations to improve span selection in NLP tasks like coreference resolution and semantic role labeling, outperforming heuristic methods.

Contribution

It presents a novel grammar-based model that replaces heuristic greedy span selection, enabling optimal span modeling using partial span annotations.

Findings

Empirical improvements on coreference resolution.

Enhanced performance on semantic role labeling.

Outperforms previous greedy span selection methods.

Abstract

Many natural language processing tasks, e.g., coreference resolution and semantic role labeling, require selecting text spans and making decisions about them. A typical approach to such tasks is to score all possible spans and greedily select spans for task-specific downstream processing. This approach, however, does not incorporate any inductive bias about what sort of spans ought to be selected, e.g., that selected spans tend to be syntactic constituents. In this paper, we propose a novel grammar-based structured span selection model which learns to make use of the partial span-level annotation provided for such problems. Compared to previous approaches, our approach gets rid of the heuristic greedy span selection scheme, allowing us to model the downstream task on an optimal set of spans. We evaluate our model on two popular span prediction tasks: coreference resolution and semantic role labeling. We show empirical improvements on both.