Improve Discourse Dependency Parsing with Contextualized Representations

TL;DR

This paper introduces a transformer-based approach to discourse dependency parsing that uses contextualized representations and sequence labeling, significantly improving performance on English and Chinese datasets.

Contribution

The paper presents a novel method leveraging transformers for dynamic, multi-level discourse unit representations and reformulates relation identification as sequence labeling.

Findings

Achieves state-of-the-art results on English datasets.

Outperforms traditional classification methods.

Effective on both English and Chinese datasets.

Abstract

Recent works show that discourse analysis benefits from modeling intra- and inter-sentential levels separately, where proper representations for text units of different granularities are desired to capture both the meaning of text units and their relations to the context. In this paper, we propose to take advantage of transformers to encode contextualized representations of units of different levels to dynamically capture the information required for discourse dependency analysis on intra- and inter-sentential levels. Motivated by the observation of writing patterns commonly shared across articles, we propose a novel method that treats discourse relation identification as a sequence labelling task, which takes advantage of structural information from the context of extracted discourse trees, and substantially outperforms traditional direct-classification methods. Experiments show that our model achieves state-of-the-art results on both English and Chinese datasets.