Guiding Extractive Summarization with Question-Answering Rewards

By Kristjan Arumae, Fei Liu. 2019

35 citations recorded by the source corpus.

Highlighting while reading is a natural behavior for people to track salient content of a document. It would be desirable to teach an extractive summarizer to do the same. However, a major obstacle to the development of a supervised summarizer is the lack of ground-truth. Manual annotation of extraction units is cost-prohibitive, whereas acquiring labels by automatically aligning human abstracts and source documents can yield inferior results. In this paper we describe a novel framework to guide a supervised, extractive summarization system with question-answering rewards. We argue that quality summaries should serve as document surrogates to answer important questions, and such question-answer pairs can be conveniently obtained from human abstracts. The system learns to promote summaries that are informative, fluent, and perform competitively on question-answering. Our results compare favorably with those reported by strong summarization baselines as evaluated by automatic metrics and human assessors.

Subfields: Natural Language Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Natural Language Processing

1. Iryna Gurevych
2. Mohit Bansal
3. Caiming Xiong
4. Lu Wang
5. Yang Gao
6. Thomas Scialom
7. Mo Yu
8. Ryan T. McDonald

Citations in the corpus

• Better Rewards Yield Better Summaries: Learning to Summarise Without References (2019, background): “Arumae and Liu (2019) propose four reward functions to train an RL-based extractive summariser,”
• On Faithfulness and Factuality in Abstractive Summarization (2020, methodology): “Finally, others have used reinforcement learning to improve informativeness and reduce contradictory information in abstractive summaries, e.g., Pasunuru and Bansal (2018) used a textual entailment-based reward and Arumae and Liu (2019), a question-answering based reward.”
• Survey of Hallucination in Natural Language Generation (2022, neutral): “Some research [3, 73, 84, 102, 120, 135, 163] adopt reinforcement learning to solve the hallucination problem, which utilizes different rewards to optimize the model.”
• Answers Unite! Unsupervised Metrics for Reinforced Summarization Models (2019, methodology): “Recent research works (Paulus et al., 2017; Pasunuru and Bansal, 2018; Arumae and Liu, 2019) have proposed to use evaluation metrics – and ROUGE in particular – to learn the model parameters through Reinforcement Learning (RL) techniques.”
• Knowledge Graph-Augmented Abstractive Summarization with Semantic-Driven Cloze Reward (2020, background): “Arumae and Liu (2019) demonstrate that using fill-in-the-blank questions by removing entities or root words leads to improved content selection.”
• Focus Attention: Promoting Faithfulness and Diversity in Summarization (2021, methodology): “We evaluate FAME models on semantic inference metrics such as textual entailment (Pasunuru and Bansal, 2018; Welleck et al., 2019b; Falke et al., 2019; Kryscinski et al., 2019) and question answering (Arumae and Liu, 2019; Wang et al., 2020a).”
• Rethinking Cooperative Rationalization: Introspective Extraction and Complement Control (2019, methodology): “It has also been applied to a relevant problem on summarization (Arumae and Liu, 2019), where the selected summary should be sufficient for answering questions related to the document.”