DiscoSum: Discourse-aware News Summarization

2025-06-07

AI Summary

News summarization often suffers from structural degradation due to language models’ limited capacity for long-range discourse modeling, compromising readability and platform-specific adaptability. To address this, we propose DiscoSum, a structure-aware multi-platform summarization framework. First, we introduce the first fine-grained news discourse structure annotation schema and a cross-platform (e.g., LinkedIn, Facebook) stylized summary dataset. Second, we design a generation mechanism integrating discourse structure modeling, structure-guided decoding, and platform-specific stylistic constraints, employing an enhanced beam search to jointly optimize narrative fidelity and stylistic customization. Experiments demonstrate that DiscoSum significantly outperforms mainstream LLM baselines in human evaluations of structural compliance, narrative coherence, and platform alignment. It also achieves superior performance on automatic metrics—including BERTScore and FactCC—validating its effectiveness in preserving factual consistency and semantic similarity.

Technology Category

• Natural Language Processing: Summarization
• Machine Learning: Large Multimodal Models (LMMs)
• Data Mining & Knowledge Management: Data Visualization & Summarization

Application Category

• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling
• Web Mining and Content Analysis: Bridging structured and unstructured data
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Recent advances in text summarization have predominantly leveraged large language models to generate concise summaries. However, language models often do not maintain long-term discourse structure, especially in news articles, where organizational flow significantly influences reader engagement. We introduce a novel approach to integrating discourse structure into summarization processes, focusing specifically on news articles across various media. We present a novel summarization dataset where news articles are summarized multiple times in different ways across different social media platforms (e.g. LinkedIn, Facebook, etc.). We develop a novel news discourse schema to describe summarization structures and a novel algorithm, DiscoSum, which employs beam search technique for structure-aware summarization, enabling the transformation of news stories to meet different stylistic and structural demands. Both human and automatic evaluation results demonstrate the efficacy of our approach in maintaining narrative fidelity and meeting structural requirements.

Problem

• Maintaining discourse structure in news summarization
• Adapting summaries for different social media platforms
• Ensuring narrative fidelity and structural requirements

Innovation

• Integrates discourse structure into summarization processes
• Develops a novel news discourse schema
• Uses beam search for structure-aware summarization