Leveraging Discourse Structure for Extractive Meeting Summarization

TL;DR

This paper presents a novel extractive meeting summarization method that uses discourse graphs and GNNs to identify key utterances, outperforming existing systems on standard datasets.

Contribution

The study introduces a discourse graph-based approach with GNNs for extractive summarization, integrating discourse structure to improve selection accuracy.

Findings

Outperforms existing text and graph-based systems on AMI and ICSI datasets

Discourse structure significantly improves summarization quality

Ablation studies highlight the importance of relation types in discourse graphs

Abstract

We introduce an extractive summarization system for meetings that leverages discourse structure to better identify salient information from complex multi-party discussions. Using discourse graphs to represent semantic relations between the contents of utterances in a meeting, we train a GNN-based node classification model to select the most important utterances, which are then combined to create an extractive summary. Experimental results on AMI and ICSI demonstrate that our approach surpasses existing text-based and graph-based extractive summarization systems, as measured by both classification and summarization metrics. Additionally, we conduct ablation studies on discourse structure and relation type to provide insights for future NLP applications leveraging discourse analysis theory.