ChainCQG: Flow-Aware Conversational Question Generation
Open accessAbstract
Conversational systems enable numerous valuable applications, and question-answering is an important component underlying many of these.However, conversational questionanswering remains challenging due to the lack of realistic, domain-specific training data.Inspired by this bottleneck, we focus on conversational question generation as a means to generate synthetic conversations for training and evaluation purposes.We present a number of novel strategies to improve conversational flow and accommodate varying question types and overall fluidity.Specifically, we design ChainCQG as a two-stage architecture that learns question-answer representations across multiple dialogue turns using a flow propagation training strategy.ChainCQG significantly outperforms both answer-aware and answer-unaware SOTA baselines (e.g., up to 48% BLEU-1 improvement).Additionally, our model is able to generate different types of questions, with improved fluidity and coreference alignment.
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In conversational systems, consistency with the conversation context and maintaining user interest are essential.
“On the other hand, in conversational systems, say, a question-asking bot, it is imperative to be consistent with the context of the conversation and also at the same time maintain the interest of the user in the conversation [2].”
From Automatic question generation: a review of methodologies, datasets, evaluation metrics, and applications (2023) · cites this paper as [2]
The [2] model outperforms several baselines, including T5 and BART-large.
“The model outperforms several baselines considered, including T5 model and BART-large.”
From Automatic question generation: a review of methodologies, datasets, evaluation metrics, and applications (2023) · cites this paper as [2]
[2] uses flow-propagation-based learning to generate conversational questions from a passage, a target answer, and prior dialogue history in a multi-turn setup.
“An architecture which employs flow-propagation-based learning for generating conversational questions is discussed in [2].”
From Automatic question generation: a review of methodologies, datasets, evaluation metrics, and applications (2023) · cites this paper as [2]
[2] uses GPT-2 to encode the answer and generate questions, and its flow-propagation training considers losses after n turns to improve conversational flow.
“For encoding the answer and for question generation, the GPT-2 model [66] is used.”
From Automatic question generation: a review of methodologies, datasets, evaluation metrics, and applications (2023) · cites this paper as [2]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
The citation lists haven’t been fetched yet
References 40 · Cited by 28
Research works printed but not linked: 3
- Philip Clarkson and Tony Robinson. 1999. Towards improved language model evaluation measures. In Sixth European Conference on Speech Communica- tion and Technology.
- Colin Raffel, Noam Shazeer, Adam Roberts, Kather- ine Lee, Sharan Narang, Michael Matena, Yanqi Zhou, Wei Li, and Peter J. Liu. 2020. Exploring the limits of transfer learning with a unified text-to- text transformer. Journal of Machine Learning Re- search, 21(140):1-67.
- Qingyang Wu, Yichi Zhang, Yu Li, and Zhou Yu. 2019. Alternating roles dialog model with large-scale pre- trained language models. arXiv, pages arXiv-1910.
- Authors
- Gu, J., Mirshekari, M., Yu, Z., & Sisto, A.
- Affiliations
- University of California, Davis (US)
- Columbia University (US)
- Venue
- Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: Main Volume
- Published
- 2021
- Pages
- 2061-2070
- OpenAlex
- W3155952370
- Cited by
- 28
- References
- 40
- Citations by year
- 2026: 3 · 2024: 2 · 2023: 16 · 2022: 6 · 2021: 1
- FWCI
- 2.89
- Type
- conference paper
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- cc-by
- Locations
- Conference of the European Chapter of the Association for Computational Linguistics (publishedVersion, cc-by)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Speech and dialogue systems
- Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
- Physical Sciences › Computer Science › Artificial Intelligence › Natural Language Processing Techniques
- Keywords
- question answering, dialogue flow, two-stage architecture
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
