DisCoDisCo at the DISRPT2021 Shared Task: A System for Discourse Segmentation, Classification, and Connective Detection

TL;DR

DisCoDisCo is a Transformer-based system that improves discourse segmentation, connective detection, and relation classification by combining contextualized embeddings with handcrafted features, achieving state-of-the-art results on the DISRPT2021 shared task.

Contribution

The paper introduces DisCoDisCo, a novel neural system that enhances Transformer embeddings with handcrafted features for discourse analysis tasks, outperforming previous models.

Findings

Outperforms SOTA on discourse segmentation and connective detection

Strong performance on the 2021 relation classification benchmark

Pre-trained models with NSP are most effective for relation classification

Abstract

This paper describes our submission to the DISRPT2021 Shared Task on Discourse Unit Segmentation, Connective Detection, and Relation Classification. Our system, called DisCoDisCo, is a Transformer-based neural classifier which enhances contextualized word embeddings (CWEs) with hand-crafted features, relying on tokenwise sequence tagging for discourse segmentation and connective detection, and a feature-rich, encoder-less sentence pair classifier for relation classification. Our results for the first two tasks outperform SOTA scores from the previous 2019 shared task, and results on relation classification suggest strong performance on the new 2021 benchmark. Ablation tests show that including features beyond CWEs are helpful for both tasks, and a partial evaluation of multiple pre-trained Transformer-based language models indicates that models pre-trained on the Next Sentence Prediction (NSP) task are optimal for relation classification.