Adapting Coreference Resolution Models through Active Learning

TL;DR

This paper investigates active learning strategies for neural coreference resolution, focusing on uncertainty sampling and annotation efficiency, to improve model transferability across domains.

Contribution

It introduces a detailed analysis of active learning methods for coreference resolution, highlighting effective span annotation strategies within documents.

Findings

Span annotation within the same document is more effective.

Uncertainty sampling improves annotation efficiency.

Error analysis guides better active learning strategies.

Abstract

Neural coreference resolution models trained on one dataset may not transfer to new, low-resource domains. Active learning mitigates this problem by sampling a small subset of data for annotators to label. While active learning is well-defined for classification tasks, its application to coreference resolution is neither well-defined nor fully understood. This paper explores how to actively label coreference, examining sources of model uncertainty and document reading costs. We compare uncertainty sampling strategies and their advantages through thorough error analysis. In both synthetic and human experiments, labeling spans within the same document is more effective than annotating spans across documents. The findings contribute to a more realistic development of coreference resolution models.