Elamur Research⌕
☰
home / papers / 2026 / Data-Efficient Adaptation to Contextual Shifts in LLM-based Conversational Recommendation

Data-Efficient Adaptation to Contextual Shifts in LLM-based Conversational Recommendation

Paper · 2026 · Hyeongjun Yang, Donghyun Kim, Seokju Hwang et al.

Summary

LLM-based conversational recommender systems (CRSs) lose quality once deployed because language use and social or cultural trends keep changing. A request that once called for one film may later call for another. Retraining on all new data is expensive, so the authors frame this as a data-efficient adaptation problem, which they say has not been formulated before. They propose CAPT (Contextual Shift-Adaptive Data Pruning and Training), which works with any underlying LLM-based CRS.

CAPT sorts new training samples into three groups:

  • familiar samples, which the old model already handles;
  • valuable samples, which are surprising but worth learning;
  • outliers, which are surprising but harmful.

A loss-based importance score, designed to stay robust when losses vary widely in scale, ranks how learnable each sample is. CAPT keeps the top-ranked samples and also uses the scores to weight the training loss and to supply harder negatives.

On three CRS benchmarks split by time, CAPT beats existing data selection methods. It equals or exceeds full-data fine-tuning while using only 10–50% of the training data, which makes periodic updates cheaper.

Key points

  • The taxonomy explains why existing selectors fall short:
    • Gradient-based methods (e.g., GREATS, LESS) filter out outliers but cannot recognise familiar samples, and they must store per-sample gradients.
    • Loss-based methods (e.g., RHO) drop familiar samples but can be dominated by outliers.
  • Shift-adaptive training has two parts:
    • Importance-based loss weighting reshapes the model's relative item preferences, so no counterfactual supervision is needed.
    • High-importance samples replace trivial random negatives in negative sampling.
  • Experiments use the REDIAL, INSPIRED, and Reddit datasets with real-world temporal splits.
  • In a LoRA setting, computing importance scores took 0.97 hours for CAPT (the same as RHO), versus 3.26 hours for GREATS and 6.85 hours for LESS. CAPT also needs no gradient storage.
  • Limitations noted by the authors:
    • CAPT targets global preference shifts, not changes within a single dialogue.
    • Long-tail items remain a common failure case.
    • Results depend on the quality of the validation set.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓Yonsei UniversitySeoul · South Korea18 works · 501 citations

 Fields, topics and concepts

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Computer Science · 18,614 worksInformation Systems · 1,730 worksDecision Sciences · 249 worksManagement Science and Operations Research · 134 works

🧭 Topics 1Research subjects and languages the work is about.Dialogue systems · 654 works

Cite

Cite as Hyeongjun Yang, Donghyun Kim, Seokju Hwang, Midan Shim, KyuHwan Yeom, KaeHyun Um and Kyong-Ho Lee. 2026. Data-Efficient Adaptation to Contextual Shifts in LLM-based Conversational Recommendation. Findings of the Association for Computational Linguistics: ACL 2026.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.findings-acl.1114

@inproceedings{yang2026data,
  title = {{Data-Efficient Adaptation to Contextual Shifts in LLM-based Conversational Recommendation}},
  author = {Hyeongjun Yang, Donghyun Kim, Seokju Hwang, Midan Shim, KyuHwan Yeom, KaeHyun Um and Kyong-Ho Lee},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  year = {2026},
  doi = {10.18653/v1/2026.findings-acl.1114},
  url = {https://doi.org/10.18653/v1/2026.findings-acl.1114}
}

References

55 references, 13 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Gediminas Adomavicius, Ramesh Sankaranarayanan, Shahana Sen, and Alexander Tuzhilin. 2005. Incorporating contextual information in recommender systems using a multidimensional approach. ACM Trans. Inf. Syst., 23(1):103–145. find
  2. Christian Bizer, Jens Lehmann, Georgi Kobilarov, Sören Auer, Christian Becker, Richard Cyganiak, and Sebastian Hellmann. 2009. Dbpedia - a crystallization point for the web of data. Web Semant., 7(3):154–165. find
  3. Lichang Chen, Shiyang Li, Jun Yan, Hai Wang, Kalpa Gunaratna, Vikas Yadav, Zheng Tang, Vijay Srinivasan, Tianyi Zhou, Heng Huang, and Hongxia Jin. 2024. Alpagasus: Training a better alpaca with fewer data. In The Twelfth International Conference on Learning Representations, ICLR 2024, Vienna, Austri…
  4. Qibin Chen, Junyang Lin, Yichang Zhang, Ming Ding, Yukuo Cen, Hongxia Yang, and Jie Tang. 2019. Towards knowledge-based recommender dialog system. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language… find
  5. Tim Dettmers, Artidoro Pagnoni, Ari Holtzman, and Luke Zettlemoyer. 2023. Qlora: Efficient finetuning of quantized llms. Advances in neural information processing systems, 36:10088–10115.
  6. Garima, Frederick Liu, Satyen Kale, and Mukund Sundararajan. 2020. Estimating training data influence by tracing gradient descent. In Proceedings of the 34th International Conference on Neural Information Processing Systems, NIPS ’20, Red Hook, NY, USA. Curran Associates Inc.
  7. Michael Gutmann and Aapo Hyvärinen. 2010. Noise-contrastive estimation: A new estimation principle for unnormalized statistical models. In Proceedings of the thirteenth international conference on artificial intelligence and statistics, pages 297–304. JMLR Workshop and Conference Proceedings. find
  8. William L. Hamilton, Jure Leskovec, and Dan Jurafsky. 2016. Cultural shift or linguistic drift? comparing two computational measures of semantic change. In Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing, pages 2116–2121, Austin, Texas. Association for Computat…
  9. Shirley Anugrah Hayati, Dongyeop Kang, Qingxi-aoyang Zhu, Weiyan Shi, and Zhou Yu. 2020. Inspired: Toward sociable recommendation dialog systems. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 8142–8152. find
  10. Zhankui He, Zhouhang Xie, Rahul Jha, Harald Steck, Dawen Liang, Yesu Feng, Bodhisattwa Prasad Majumder, Nathan Kallus, and Julian McAuley. 2023. Large language models as zero-shot conversational recommenders. In Proceedings of the 32nd ACM international conference on information and knowledge manage… find
  11. Zhankui He, Zhouhang Xie, Harald Steck, Dawen Liang, Rahul Jha, Nathan Kallus, and Julian McAuley. 2025. Reindex-then-adapt: Improving large language models for conversational recommendation. In Proceedings of the Eighteenth ACM International Conference on Web Search and Data Mining, WSDM ’25, page… find
  12. Edward J. Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. 2022. Lora: Low-rank adaptation of large language models. In The Tenth International Conference on Learning Representations, ICLR 2022, Virtual Event, April 25-29, 2022.
  13. Dietmar Jannach, Ahtsham Manzoor, Wanling Cai, and Li Chen. 2021. A survey on conversational recommender systems. ACM Comput. Surv., 54(5). find
  14. Angela H Jiang, Daniel L-K Wong, Giulio Zhou, David G Andersen, Jeffrey Dean, Gregory R Ganger, Gauri Joshi, Michael Kaminksy, Michael Kozuch, Zachary C Lipton, and 1 others. 2019. Accelerating deep learning by focusing on the biggest losers. arXiv preprint arXiv:1910.00762. — [arXiv:1910.00762](https://arxiv.org/abs/1910.00762) find
  15. Xisen Jin, Dejiao Zhang, Henghui Zhu, Wei Xiao, Shang-Wen Li, Xiaokai Wei, Andrew Arnold, and Xiang Ren. 2022. Lifelong pretraining: Continually adapting language models to emerging corpora. In Proceedings of BigScience Episode \#5 – Workshop on Challenges & Perspectives in Creating Large Language Mo… find
  16. Krishnateja Killamsetty, Durga S, Ganesh Ramakrishnan, Abir De, and Rishabh Iyer. 2021. Grad-match: Gradient matching based data subset selection for efficient deep model training. In Proceedings of the 38th International Conference on Machine Learning, volume 139 of Proceedings of Machine Learning… find
  17. Pang Wei Koh and Percy Liang. 2017. Understanding black-box predictions via influence functions. In Proceedings of the 34th International Conference on Machine Learning - Volume 70, ICML’17, page 1885–1894. JMLR.org.
  18. Yehuda Koren. 2009. Collaborative filtering with temporal dynamics. In Proceedings of the 15th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD ’09, page 447–456, New York, NY, USA. Association for Computing Machinery. find
  19. Angeliki Lazaridou, Adhiguna Kuncoro, Elena Gribovskaya, Devang Agrawal, Adam Liška, Tayfun Terzi, Mai Gimenez, Cyprien de Masson d’Autume, Tomas Kocisky, Sebastian Ruder, Dani Yogatama, Kris Cao, Susannah Young, and Phil Blunsom. 2021. Mind the gap: assessing temporal generalization in neural langu… find
  20. Chuang Li, Yang Deng, Hengchang Hu, Min-Yen Kan, and Haizhou Li. 2025. ChatCRS: Incorporating external knowledge and goal guidance for LLM-based conversational recommender systems. In Findings of the Association for Computational Linguistics: NAACL 2025, pages 295–312, Albuquerque, New Mexico. Assoc… find
  21. Raymond Li, Samira Ebrahimi Kahou, Hannes Schulz, Vincent Michalski, Laurent Charlin, and Chris Pal. 2018. Towards deep conversational recommendations. Advances in neural information processing systems, 31. find
  22. Xiang Lisa Li and Percy Liang. 2021. Prefix-tuning: Optimizing continuous prompts for generation. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers), pages 4582–… find
  23. Dongding Lin, Jian Wang, Wenjie Li, abc, and abc. 2023. Cola: Improving conversational recommender systems by collaborative augmentation. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 37, pages 4462–4470. find
  24. Xinyu Lin, Wenjie Wang, Yongqi Li, Shuo Yang, Fuli Feng, Yinwei Wei, and Tat-Seng Chua. 2024. Data-efficient fine-tuning for llm-based recommendation. In Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval, SIGIR ’24, page 365–374, New York… find
  25. Ilya Loshchilov and Frank Hutter. 2016. Online batch selection for faster training of neural networks. International Conference on Learning Representations Workshop. find
  26. Daniel Loureiro, Francesco Barbieri, Leonardo Neves, Luis Espinosa Anke, and Jose Camacho-collados. 2022. TimeLMs: Diachronic language models from Twitter. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics: System Demonstrations, pages 251–260, Dublin, Irelan…
  27. Kelvin Luu, Daniel Khashabi, Suchin Gururangan, Karishma Mandyam, and Noah A. Smith. 2022. Time waits for no one! analysis and challenges of temporal misalignment. In Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Tec…
  28. Wenchang Ma, Ryuichi Takanobu, Minlie Huang, and abc. 2021. Cr-walker: Tree-structured graph reasoning and dialog acts for conversational recommendation. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 1839–1851. find
  29. Kevin Meng, Arnab Sen Sharma, Alex J. Andonian, Yonatan Belinkov, and David Bau. 2023. Mass-editing memory in a transformer. In The Eleventh International Conference on Learning Representations, ICLR 2023, Kigali, Rwanda, May 1-5, 2023. find
  30. Meryem M’hamdi and Jonathan May. 2024. Leitner-guided memory replay for cross-lingual continual learning. In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers), pages 7808–7821, Mexico… find
  31. Sören Mindermann, Jan M Brauner, Muhammed T Razzak, Mrinank Sharma, Andreas Kirsch, Winnie Xu, Benedikt Höltgen, Aidan N Gomez, Adrien Morisot, Sebastian Farquhar, and Yarin Gal. 2022. Prioritized training on points that are learnable, worth learning, and not yet learnt. In Proceedings of the 39th I… find
  32. Baharan Mirzasoleiman, Jeff Bilmes, and Jure Leskovec. 2020. Coresets for data-efficient training of machine learning models. In Proceedings of the 37th International Conference on Machine Learning, ICML’20. JMLR.org. find
  33. Eric Mitchell, Charles Lin, Antoine Bosselut, Chelsea Finn, and Christopher D. Manning. 2022. Fast model editing at scale. In The Tenth International Conference on Learning Representations, ICLR 2022, Virtual Event, April 25-29, 2022. find
  34. Niklas Muennighoff, SU Hongjin, Liang Wang, Nan Yang, Furu Wei, Tao Yu, Amanpreet Singh, and Douwe Kiela. Generative representational instruction tuning. In The Thirteenth International Conference on Learning Representations. Sung Min Park, Kristian Georgiev, Andrew Ilyas, Guillaume Leclerc, and Ale…
  35. Mansheej Paul, Surya Ganguli, and Gintare Karolina Dziugaite. 2021. Deep learning on a data diet: finding important examples early in training. In Proceedings of the 35th International Conference on Neural Information Processing Systems, NIPS ’21, Red Hook, NY, USA. Curran Associates Inc. find
  36. Mathieu Ravaut, Hao Zhang, Lu Xu, Aixin Sun, and Yong Liu. 2024. Parameter-efficient conversational recommender system as a language processing task. In Proceedings of the 18th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers), pages 152–165. find
  37. Paul Röttger and Janet Pierrehumbert. 2021. Temporal adaptation of BERT and performance on downstream document classification: Insights from social media. In Findings of the Association for Computational Linguistics: EMNLP 2021, pages 2400–2412, Punta Cana, Dominican Republic. Association for Comput… find
  38. Ayrton San Joaquin, Bin Wang, Zhengyuan Liu, Nicholas Asher, Brian Lim, Philippe Muller, and Nancy F. Chen. 2024. In2Core: Leveraging influence functions for coreset selection in instruction finetuning of large language models. In Findings of the Association for Computational Linguistics: EMNLP 2024… find
  39. Jiachen Tianhao Wang, Tong Wu, Dawn Song, Prateek Mittal, and Ruoxi Jia. 2024a. Greats: Online selection of high-quality data for llm training in every iteration. Advances in Neural Information Processing Systems, 37:131197–131223. find
  40. Peng Wang, Zexi Li, Ningyu Zhang, Ziwen Xu, Yunzhi Yao, Yong Jiang, Pengjun Xie, Fei Huang, and Huajun Chen. 2025. Wise: rethinking the knowledge memory for lifelong model editing of large language models. In Proceedings of the 38th International Conference on Neural Information Processing Systems,… find
  41. Yifan Wang, Yafei Liu, Chufan Shi, Haoling Li, Chen Chen, Haonan Lu, and Yujiu Yang. 2024b. InsCL: A data-efficient continual learning paradigm for fine-tuning large language models with instructions. In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computat… find
  42. Alexander Wettig, Aatmik Gupta, Saumya Malik, and Danqi Chen. 2024. Qurating: selecting high-quality data for training language models. In Proceedings of the 41st International Conference on Machine Learning, ICML’24. JMLR.org.
  43. Heesoo Won, Byungkook Oh, Hyeongjun Yang, and Kyong-Ho Lee. 2023. Cross-modal contrastive learning for aspect-based recommendation. Information Fusion, 99:101858. find
  44. Tongtong Wu, Linhao Luo, Yuan-Fang Li, Shirui Pan, Thuy-Trang Vu, and Gholamreza Haffari. 2024. Continual learning for large language models: A survey. Preprint, arXiv:2402.01364.
  45. Yunjia Xi, Weiwen Liu, Jianghao Lin, Bo Chen, Ruiming Tang, Weinan Zhang, and Yong Yu. 2024. Mem-ocrs: Memory-enhanced sequential conversational recommender systems with large language models. In Proceedings of the 33rd ACM International Conference on Information and Knowledge Management, CIKM ’24,… find
  46. Hyeongjun Yang, Heesoo Won, Youbin Ahn, and Kyong-Ho Lee. 2023. Click: Contrastive learning for injecting contextual knowledge to conversational recommender system. In Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics, pages 1867–1877. find
  47. Ting Yang and Li Chen. 2024. Unleashing the retrieval potential of large language models in conversational recommender systems. In Proceedings of the 18th ACM Conference on Recommender Systems, RecSys ’24, page 43–52, New York, NY, USA. Association for Computing Machinery. find
  48. Bowen Zhao, Zander Brumbaugh, Yizhong Wang, Hannaneh Hajishirzi, and Noah Smith. 2024. Set the clock: Temporal alignment of pretrained language models. In Findings of the Association for Computational Linguistics: ACL 2024, pages 15015–15040, Bangkok, Thailand. Association for Computational Linguist… find
  49. Haizhong Zheng, Rui Liu, Fan Lai, and Atul Prakash. 2023. Coverage-centric coreset selection for high pruning rates. In The Eleventh International Conference on Learning Representations, ICLR 2023, Kigali, Rwanda, May 1-5, 2023. find
  50. Junhao Zheng, Shengjie Qiu, Chengming Shi, and Qianli Ma. 2025. Towards lifelong learning of large language models: A survey. ACM Comput. Surv., 57(8).
  51. Kun Zhou, Wayne Xin Zhao, Shuqing Bian, Yuanhang Zhou, Ji-Rong Wen, and Jingsong Yu. 2020. Improving conversational recommender systems via knowledge graph based semantic fusion. In Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, pages 1006– 1014. find
  52. Yuanhang Zhou, Kun Zhou, Wayne Xin Zhao, Cheng Wang, Peng Jiang, and He Hu. 2022. C 2 -crs: Coarse-to-fine contrastive learning for conversational recommender system. In Proceedings of the Fifteenth ACM International Conference on Web Search and Data Mining, pages 1488–1496. find
  53. Mengzhou Xia, Sadhika Malladi, Suchin Gururangan, Sanjeev Arora, and Danqi Chen. 2024. Less: selecting influential data for targeted instruction tuning. In Proceedings of the 41st International Conference on Machine Learning, ICML’24. JMLR.org.
  54. Hyeongjun Yang, Donghyun Kim, Gayeon Park, KyuHwan Yeom, and Kyong-Ho Lee. 2025. Coresense: Social commonsense knowledge-aware context refinement for conversational recommender system. IEEE Transactions on Knowledge and Data Engineering, 37(4):1702–1713. find
  55. Hyeongjun Yang, Yerim Lee, Gayeon Park, Taeyoung Kim, Heesun Kim, Kyong-Ho Lee, and Byungkook Oh. 2024. Granular intents learning via mutual information maximization for knowledge-aware recommendation. Knowledge-Based Systems, 306:112705. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/yang-2026-data-efficient-adaptation-contextual-shifts/