Anaphora Resolution in Dialogue: System Description (CODI-CRAC 2022 Shared Task)

TL;DR

This paper describes models for anaphora resolution in dialogue, combining clustering, coreference, and discourse deixis techniques, achieving significant improvements in shared task performance.

Contribution

It introduces a hybrid approach integrating incremental clustering with coreference models and multi-task learning for discourse deixis resolution.

Findings

Up to 10.33% improvement over baseline clustering.

Effective multi-task learning for discourse deixis.

Adaptation of higher-order resolution model for bridging.

Abstract

We describe three models submitted for the CODI-CRAC 2022 shared task. To perform identity anaphora resolution, we test several combinations of the incremental clustering approach based on the Workspace Coreference System (WCS) with other coreference models. The best result is achieved by adding the ''cluster merging'' version of the coref-hoi model, which brings up to 10.33% improvement 1 over vanilla WCS clustering. Discourse deixis resolution is implemented as multi-task learning: we combine the learning objective of corefhoi with anaphor type classification. We adapt the higher-order resolution model introduced in Joshi et al. (2019) for bridging resolution given gold mentions and anaphors.