Top-down Discourse Parsing via Sequence Labelling

TL;DR

This paper presents a simplified top-down discourse parsing method using sequence labelling, leveraging modern models and a novel dynamic oracle to improve RST parsing performance.

Contribution

It introduces a top-down discourse parsing approach as sequence labelling, eliminating the decoder and reducing search complexity, with a new dynamic oracle and state-of-the-art results.

Findings

LSTM model achieves new state-of-the-art in RST parsing

Sequence labelling simplifies discourse parsing process

Dynamic oracle enhances top-down parsing accuracy

Abstract

We introduce a top-down approach to discourse parsing that is conceptually simpler than its predecessors (Kobayashi et al., 2020; Zhang et al., 2020). By framing the task as a sequence labelling problem where the goal is to iteratively segment a document into individual discourse units, we are able to eliminate the decoder and reduce the search space for splitting points. We explore both traditional recurrent models and modern pre-trained transformer models for the task, and additionally introduce a novel dynamic oracle for top-down parsing. Based on the Full metric, our proposed LSTM model sets a new state-of-the-art for RST parsing.