Abstract:Nowadays, neural text generation has made tremendous progress in abstractive summarization tasks. However, most of the existing summarization models take in the whole document all at once, which sometimes cannot meet the needs in practice. Practically, social text streams such as news events and tweets keep growing from time to time, and can only be fed to the summarization system step by step. Hence, in this paper, we propose the task of Stepwise Summarization, which aims to generate a new appended summary each time a new document is proposed. The appended summary should not only summarize the newly added content but also be coherent with the previous summary, to form an up-to-date complete summary. To tackle this challenge, we design an adversarial learning model, named Stepwise Summary Generator (SSG). First, SSG selectively processes the new document under the guidance of the previous summary, obtaining polished document representation. Next, SSG generates the summary considering both the previous summary and the document. Finally, a convolutional-based discriminator is employed to determine whether the newly generated summary is coherent with the previous summary. For the experiment, we extend the traditional two-step update summarization setting to a multi-step stepwise setting, and re-propose a large-scale stepwise summarization dataset based on a public story generation dataset. Extensive experiments on this dataset show that SSG achieves state-of-the-art performance in terms of both automatic metrics and human evaluations. Ablation studies demonstrate the effectiveness of each module in our framework. We also discuss the benefits and limitations of recent large language models on this task.

A Two-Stage Long Text Summarization Method Based on Discourse Structure

Efficient Two-stage Approach for Long Document Summarization

Summarize, Outline, and Elaborate: Long-Text Generation via Hierarchical Supervision from Extractive Summaries

A New Approach to Overgenerating and Scoring Abstractive Summaries

Incorporating Distributions of Discourse Structure for Long Document Abstractive Summarization

GoSum: Extractive Summarization of Long Documents by Reinforcement Learning and Graph Organized discourse state

Discourse-Aware Unsupervised Summarization of Long Scientific Documents

An Unsupervised Extractive Summarization Method Based on Multi-Round Computation

Dual-Level Contrastive Learning for Improving Conciseness of Summarization

Abstractive Summarization Improved by WordNet-based Extractive Sentences

EDU-level Extractive Summarization with Varying Summary Lengths

Summ^N: A Multi-Stage Summarization Framework for Long Input Dialogues and Documents

A study of extractive summarization of long documents incorporating local topic and hierarchical information

Improving Multi-Stage Long Document Summarization with Enhanced Coarse Summarizer

A Syntax-Augmented and Headline-Aware Neural Text Summarization Method

Write Summary Step-by-Step: A Pilot Study of Stepwise Summarization

Automatic Text Summarization Method Based on Improved TextRank Algorithm and K-Means Clustering

SgSum: Transforming Multi-document Summarization into Sub-graph Selection

Composing Elementary Discourse Units in Abstractive Summarization.

Extractive Summarization As Text Matching

DCDSum: An interpretable extractive summarization framework based on contrastive learning method