Make The Most of Prior Data: A Solution for Interactive Text Summarization with Preference Feedback
Open accessAbstract
For summarization, human preferences is critical to tame outputs of the summarizer in favor of human interests, as ground-truth summaries are scarce and ambiguous.Practical settings require dynamic exchanges between humans and AI agents wherein feedback is provided in an online manner, a few at a time.In this paper, we introduce a new framework to train summarization models with preference feedback interactively.By properly leveraging offline data and a novel reward model, we improve the performance regarding ROUGE scores and sample-efficiency.Our experiments on three various datasets confirm the benefit of the proposed framework in active, few-shot and online settings of preference learning.
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?
Learning algorithms such as PPO in training sequence-to-sequence models struggle to improve from poor initializations, especially when the action space is large, as in text generation.
“From an RL perspective, learning algorithms such as the widely used PPO in training sequence-to-sequence models, struggle to improve from poor initializations, especially when the action space is large, as in the case of text generation.”
From RLHF Deciphered: A Critical Analysis of Reinforcement Learning from Human Feedback for LLMs (2026) · cites this paper as [128]
Section 7.2 reports that a baseline trained with PPO could not improve the summarization agent with only a few interactions. But Section 3.2 describes a supervised-initialized model choosing between two sentence labels, not a sequence-to-sequence text generator with a large action space. The paper does not attribute the baseline’s difficulty to poor initialization or action-space size; page numbers are unavailable in our copy. (LLM, conf=0.86)
Experimental — accuracy not yet measured
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Cites
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2019) · Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers) · doi:10.18653/v1/n19-1423 · Cited by 48,293 · see citing works
Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks
Reimers, N., & Gurevych, I. (2019) · Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP) · doi:10.18653/v1/d19-1410 · Cited by 12,778 · see citing works
Proximal Policy Optimization Algorithms
Schulman, J., Wolski, F., Dhariwal, P., Radford, A., & Klimov, O. (2017) · arXiv · doi:10.48550/arxiv.1707.06347 · Cited by 10,625 · see citing works
(2004) · Cited by 9,718 · see citing works
Rank Analysis of Incomplete Block Designs: I. The Method of Paired Comparisons
Bradley, R. A., & Terry, M. E. (1952) · Biometrika · doi:10.2307/2334029 · Cited by 2,412 · see citing works
High-Dimensional Continuous Control Using Generalized Advantage Estimation
Schulman, J., Moritz, P., Levine, S., Jordan, M., & Abbeel, P. (2015) · arXiv · doi:10.48550/arxiv.1506.02438 · Cited by 2,103 · see citing works
Text Summarization with Pretrained Encoders
Liu, Y., & Lapata, M. (2019) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/d19-1387 · Cited by 1,779 · see citing works
A Deep Reinforced Model for Abstractive Summarization
Paulus, R., Xiong, C., & Socher, R. (2017) · arXiv · doi:10.48550/arxiv.1705.04304 · Cited by 1,182 · see citing works
A survey of human-in-the-loop for machine learning
Wu, X., Xiao, L., Sun, Y., Zhang, J., Ma, T., & He, L. (2022) · Future Generation Computer Systems · doi:10.1016/j.future.2022.05.014 · Cited by 800 · see citing works
Learning to summarize from human feedback
Stiennon, N., Ouyang, L., Wu, J., Ziegler, D. M., Lowe, R., Voss, C., Radford, A., Amodei, D., & Christiano, P. (2020) · arXiv · doi:10.48550/arxiv.2009.01325 · Cited by 550 · see citing works
Fine-Tuning Language Models from Human Preferences
Ziegler, D. M., Stiennon, N., Wu, J., Brown, T. B., Radford, A., Amodei, D., Christiano, P., & Irving, G. (2019) · arXiv · doi:10.48550/arxiv.1909.08593 · Cited by 413 · see citing works
Viewpoint: Human-in-the-loop Artificial Intelligence
Zanzotto, F. M. (2019) · Journal of Artificial Intelligence Research · doi:10.1613/jair.1.11345 · Cited by 380 · see citing works
BillSum: A Corpus for Automatic Summarization of US Legislation
Kornilova, A., & Eidelman, V. (2019) · Conference on Empirical Methods in Natural Language Processing · doi:10.18653/v1/d19-5406 · Cited by 232 · see citing works
Siddhant, A., & Lipton, Z. C. (2018) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/d18-1318 · Cited by 182 · see citing works
Survey on reinforcement learning for language processing
Uc-Cetina, V., Navarro-Guerrero, N., Martin-Gonzalez, A., Weber, C., & Wermter, S. (2023) · Artificial Intelligence Review · doi:10.1007/s10462-022-10205-5 · Cited by 177 · see citing works
Multi-Reward Reinforced Summarization with Saliency and Entailment
Pasunuru, R., & Bansal, M. (2018) · Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT) · doi:10.18653/v1/n18-2102 · Cited by 166 · see citing works
Correcting the Triplet Selection Bias for Triplet Loss
Yu, B., Liu, T., Gong, M., Ding, C., & Tao, D. (2018) · Lecture notes in computer science · doi:10.1007/978-3-030-01231-1_5 · Cited by 106 · see citing works
Self-Supervised Contextual Keyword and Keyphrase Retrieval with Self-LabellingPreprint
Sharma, P., & Li, Y. (2019) · Preprints.org · doi:10.20944/preprints201908.0073.v1 · Cited by 101 · see citing works
Abstractive Summarization of Reddit Posts with Multi-level Memory Networks
Kim, B., Kim, H., & Kim, G. (2019) · Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers) · doi:10.18653/v1/n19-1260 · Cited by 96 · see citing works
A survey of preference-based reinforcement learning methods
WirthChristian, AkrourRiad, NeumannGerhard, & FürnkranzJohannes (2017) · Journal of Machine Learning Research · doi:10.5555/3122009.3208017 · Cited by 94 · see citing works
Deep Reinforcement Learning with Distributional Semantic Rewards for Abstractive Summarization
Li, S., Lei, D., Qin, P., & Wang, W. Y. (2019) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/d19-1623 · Cited by 40 · see citing works
Pvs, A., & Meyer, C. M. (2017) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/p17-1124 · Cited by 35 · see citing works
Gao, Y., Meyer, C. M., & Gurevych, I. (2018) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/d18-1445 · Cited by 33 · see citing works
Robust Deep Reinforcement Learning for Extractive Legal Summarization
Nguyen, D.-H., Nguyen, B.-S., Nghiem, N. V. D., Le, D. T., Khatun, M. A., Nguyen, M.-T., & Lê, H. (2021) · Communications in computer and information science · doi:10.1007/978-3-030-92310-5_69 · Cited by 26 · see citing works
Self-citationThe Rule of Three: Abstractive Text Summarization in Three Bullet PointsPreprint
Kodaira, T., & Komachi, M. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1809.10867 · Cited by 2 · see citing works
Flagged first, then by citations
1 of the 26 references couldn’t be matched to a known paper.
Research works printed but not linked: 2
- David Maxwell, Leif Azzopardi, and Yashar Mosh- feghi. 2017. A study of snippet length and informa- tiveness: Behaviour, performance and user experi- ence. In Proceedings of the 40th International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 135-144.
- Shashi Narayan, Shay B. Cohen, and Mirella Lapata. 2018. Ranking sentences for extractive summariza- tion with reinforcement learning. In Proceedings of the 2018 Conference of the North American Chap- ter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Pa- pers), pages 1747-1759, New Orleans, Louisiana. Association for Computational Linguistics. · DOI
Cited by
Sorted by citations
RLHF Deciphered: A Critical Analysis of Reinforcement Learning from Human Feedback for LLMs
Chaudhari, S., Aggarwal, P., Murahari, V., Rajpurohit, T., Kalyan, A., Narasimhan, K., Deshpande, A., & Castro da Silva, B. (2026) · ACM Computing Surveys · doi:10.1145/3743127 · Cited by 92 · see citing works
Bridging the Gap: A Survey on Integrating (Human) Feedback for Natural Language Generation
Fernandes, P., Madaan, A., Liu, E., Farinhas, A., Martins, P. H., Bertsch, A., de Souza, J. G. C., Zhou, S., Wu, T., Neubig, G., et al. (2023) · Transactions of the Association for Computational Linguistics · doi:10.1162/tacl_a_00626 · Cited by 57 · see citing works
Kirk, H. R., Bean, A., Vidgen, B., Röttger, P., & Hale, S. A. (2023) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2023.emnlp-main.148 · Cited by 26 · see citing works
On Improving Summarization Factual Consistency from Natural Language Feedback
Liu, Y., Deb, B., Teruel, M., Halfaker, A., Radev, D., & Awadallah, A. H. (2023) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2023.acl-long.844 · Cited by 17 · see citing works
SPEC: Summary Preference Decomposition for Low-Resource Abstractive Summarization
Chen, Y.-S., Song, Y.-Z., & Shuai, H.-H. (2022) · IEEE/ACM Transactions on Audio Speech and Language Processing · doi:10.1109/taslp.2022.3230539 · Cited by 8 · see citing works
Improving Document Image Understanding with Reinforcement Finetuning
Nguyen, B.-S., Le, D. T., Vu, H. M., Nguyen, T.-A., Nguyen, M.-T., & Lê, H. (2023) · Communications in computer and information science · doi:10.1007/978-981-99-1648-1_5 · Cited by 1 · see citing works
Self-citation
- Authors
- Nguyen, D.-H., Nghiem, N. V. D., Nguyen, B.-S., Tien Le, D. T., Sabahi, S., Nguyen, M.-T., & Le, H.
- Affiliations
- Hung Yen University of Technology and Education (VN)
- Venue
- Findings of the Association for Computational Linguistics: NAACL 2022
- Published
- 2022
- Pages
- 1919-1930
- OpenAlex
- W4223477065
- Cited by
- 6
- References
- 26
- Citations by year
- 2025: 1 · 2023: 4 · 2022: 1
- FWCI
- 0.83
- Type
- conference paper
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- cc-by
- Locations
- Findings of the Association for Computational Linguistics: NAACL (publishedVersion, cc-by)
- arXiv (Cornell University) (submittedVersion)
- Figshare (submittedVersion, cc-by)
- arXiv (Cornell University)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
- Physical Sciences › Computer Science › Artificial Intelligence › Advanced Text Analysis Techniques
- Physical Sciences › Computer Science › Artificial Intelligence › Natural Language Processing Techniques
- Keywords
- preference feedback, preference learning, reward modeling, online feedback, sample efficiency, human-AI interaction
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
