Harvesting Paragraph-level Question-Answer Pairs from Wikipedia
By Xinya Du, Claire Cardie. 2018
174 citations recorded by the source corpus.
We study the task of generating from Wikipedia articles question-answer pairs that cover content beyond a single sentence. We propose a neural network approach that incorporates coreference knowledge via a novel gating mechanism. As compared to models that only take into account sentence-level information (Heilman and Smith, 2010; Du et al., 2017; Zhou et al., 2017), we find that the linguistic knowledge introduced by the coreference representation aids question generation significantly, producing models that outperform the current state-of-the-art. We apply our system (composed of an answer span extraction system and the passage-level QG system) to the 10,000 top ranking Wikipedia articles and create a corpus of over one million question-answer pairs. We provide qualitative analysis for the this large-scale generated corpus from Wikipedia.
Subfields: Artificial Intelligence, Communication, Retrieval-Augmented Generation, 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 Artificial Intelligence
- Mohit Bansal
- Jianfeng Gao
- Luke S. Zettlemoyer
- Yejin Choi
- Furu Wei
- Hannaneh Hajishirzi
- Percy Liang
- Graham Neubig
In Communication
In Retrieval-Augmented Generation
- Luke S. Zettlemoyer
- Sebastian Riedel
- Douwe Kiela
- Jianfeng Gao
- Michael Lewis
- Hangbo Bao
- Mohit Bansal
- Nan Yang
In Natural Language Processing
- Mohit Bansal
- Yejin Choi
- Luke S. Zettlemoyer
- Jianfeng Gao
- Furu Wei
- Hannaneh Hajishirzi
- Percy Liang
- Douwe Kiela
Citations in the corpus
- Addressing Semantic Drift in Question Generation for Semi-Supervised Question Answering (2019, methodology): “unlabeled paragraph and answer labels from HarvestingQA (Du and Cardie, 2018), and have different QG systems to label questions.”
- Models in the Loop: Aiding Crowdworkers with Generative Annotation Assistants (2021, background): “A long line of prior work has trained generative models for question answering (Du et al., 2017; Du and Cardie, 2018; Zhao et al., 2018; Lewis and Fan, 2019; Alberti et al., 2019; Puri et al., 2020; Yang et al., 2020; Bartolo et al., 2021; Lewis et al., 2021).”
- MiniLM: Deep Self-Attention Distillation for Task-Agnostic Compression of Pre-Trained Transformers (2020, methodology): “We conduct experiments for the answer-aware question generation task (Du & Cardie, 2018).”
- UniLMv2: Pseudo-Masked Language Models for Unified Language Model Pre-Training (2020, methodology): “We perform evaluations on question generation (Du & Cardie, 2018), the task of automatically producing relevant questions that ask for the given answer and context.”
- Unsupervised Commonsense Question Answering with Self-Talk (2020, background): “Motivations vary from data augmentation to QA tasks (Du et al., 2017; Dhingra et al., 2018; Du and Cardie, 2018; Sachan and Xing, 2018; Fabbri et al., 2020) through conversational machine read-”
- Synthetic QA Corpora Generation with Roundtrip Consistency (2019, background): “Question generation is a well-studied task in its own right (Heilman and Smith, 2010; Du et al., 2017; Du and Cardie, 2018).”
- Cross-Lingual Natural Language Generation via Pre-Training (2019, neutral): “We compare our model to the following baselines: • C OREF N QG (Du and Cardie 2018) An attentional sequence-to-sequence model with a feature-rich encoder.”
- Unsupervised Question Answering by Cloze Translation (2019, methodology): “Also, whilst supervised models exist for this task, they require the type of annotation unavailable in this setting (Du et al. 2017; Du and Cardie 2018; Hosking and Riedel 2019, inter alia ).”
- Improving Question Answering Model Robustness with Synthetic Adversarial Data Generation (2021, background): “Many approaches have been proposed to generate question-answer pairs given a passage (Du et al., 2017; Du and Cardie, 2018; Zhao et al., 2018; Lewis and Fan, 2019; Alberti et al., 2019; Puri et al., 2020; Lewis et al., 2021b).”
- Question Answering Infused Pre-training of General-Purpose Contextualized Representations (2021, background): “Neural question generation has been well-studied for different purposes (Du et al., 2017; Du and Cardie, 2018; Zhao et al., 2018; Lewis and Fan, 2019; Alberti et al., 2019; Puri et al., 2020; Lewis et al., 2021; Bartolo et al., 2021).”
- Review-based Question Generation with Adaptive Instance Transfer and Augmentation (2019, background): “, Wikipedia (Du and Cardie, 2018), most of the existing QG works treat it as a dual task of reading”
- Quiz-Style Question Generation for News Stories (2021, unknown): “There have also been various extensions, including copy or pointer mechanisms to overcome out-of-vocabulary issues [11, 44], mechanisms for handling paragraph-length inputs [50, 51], and more [19, 33].”
- Synthetic Data Augmentation for Zero-Shot Cross-Lingual Question Answering (2020, background): “On English corpora, generating synthetic questions has shown to significantly improve the performance of QA models (Du et al., 2017; Golub et al., 2017; Du and Cardie, 2018; Alberti et al., 2019).”
- Transformer-based End-to-End Question Generation (2020, methodology): “Our model outperforms prior RNN-based Seq2Seq works (Du et al., 2017; Du and Cardie, 2018; Zhao et al., 2018) in terms of METEOR and ROUGE L score.”
- Learning with Limited Data for Multilingual Reading Comprehension (2019, methodology): “Second, automatic labeling function (Du and Cardie, 2018; Serban et al., 2016) can be adopted to generate synthetic labels in a target language.”
- Supplementary Material: Deep Self-Attention Distillation for Task-Agnostic Compression of Pre-Trained Transformers (2020, methodology): “#Param BLEU-4 METEOR ROUGE-L Du and Cardie [7] 15.”
- Learning to Generate Questions by LearningWhat not to Generate (2019, background): “To resolve this issue, recent works assume the aspect is known when generating a question [16, 31, 32, 38, 42] or can be detected by a third-party pipeline [7].”