Elamur Research⌕
☰
home / papers / 2026 / What Makes LLMs Effective Sequential Recommenders? A Study on Preference Intensity and Temporal Context

What Makes LLMs Effective Sequential Recommenders? A Study on Preference Intensity and Temporal Context

Paper · 2026 · Zhongyu Ouyang, Qianlong Wen, Chunhui Zhang et al.

Summary

LLM-based sequential recommenders are usually tuned with DPO-style alignment, which reduces user feedback to simple preferred-versus-dispreferred pairs. The authors ask which properties of preference data actually help an LLM predict the next item a user will like. They argue that two signals are lost: how strongly a user likes or dislikes an item (preference intensity), and how close an interaction is to the current decision (temporal context). In a proof-of-concept study, they fine-tune LLaMA3-8B on MovieLens and Amazon-Books. Keeping negatively rated items in the history helps only when ratings accompany them, and the best accuracy comes from combining both. Building on this, they introduce RecPO. It turns explicit and implicit feedback into a shared preference score and combines that score with recency to set adaptive reward margins, instead of using one uniform margin. Across five datasets, RecPO outperforms state-of-the-art baselines such as S-DPO. It also behaves in human-like ways, for example by steering away from items users dislike. The study points toward richer, time-aware supervision for LLM recommenders.

Key points

  • The pilot experiment compares four history setups (Filtered Items, Full Items, each with or without ratings) using SFT on LLaMA3-8B, evaluated with Hit Ratio@1. Unannotated negative items act as noise, while rated negatives improve results.
  • RecPO extends S-DPO-style multi-negative alignment with margins that depend on both rating strength (e.g., 5-star vs. 3-star) and interaction recency.
  • RecPO identifies users' least-preferred items more accurately than SFT and S-DPO ("aversion accuracy"), even though it receives no explicit aversion labels.
  • On BeerAdvocate, RecPO stays stable across different history lengths: the variance of its Hit Ratio@1 is 8.7%, versus 17.8% for S-DPO.
  • The margin weight λ is set to 2 by default. Larger values raise Hit Ratio@1 but lower Valid Ratio, which measures how well the model follows instructions. Code is available at github.com/zyouyang/RecPO.

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

🎓University of Notre DameNotre Dame · United States51 works · 36,717 citations🎓Dartmouth CollegeHanover · United States37 works · 42,625 citations

 Fields, topics and concepts

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksInformation Systems · 1,437 worksDecision Sciences · 209 worksManagement Science and Operations Research · 108 works

🧭 Topics 1Research subjects and languages the work is about.Language models & pretraining · 3,320 works

Cite

Cite as Zhongyu Ouyang, Qianlong Wen, Chunhui Zhang, Yanfang Ye and Soroush Vosoughi. 2026. What Makes LLMs Effective Sequential Recommenders? A Study on Preference Intensity and Temporal Context. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.acl-long.656

@inproceedings{ouyang2026what,
  title = {{What Makes LLMs Effective Sequential Recommenders? A Study on Preference Intensity and Temporal Context}},
  author = {Zhongyu Ouyang, Qianlong Wen, Chunhui Zhang, Yanfang Ye and Soroush Vosoughi},
  booktitle = {Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2026},
  doi = {10.18653/v1/2026.acl-long.656},
  url = {https://doi.org/10.18653/v1/2026.acl-long.656}
}

References

61 references, 12 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Afra Amini, Tim Vieira, and Ryan Cotterell. 2024. Direct preference optimization with an offset. In Findings of the ACL. find
  2. Janet Wilde Astington and Jennifer M Jenkins. 1995. Theory of mind development and social understanding. Cognition & Emotion. find
  3. Mohammad Gheshlaghi Azar, Zhaohan Daniel Guo, Bilal Piot, Remi Munos, Mark Rowland, Michal Valko, and Daniele Calandriello. 2024. A general theoretical paradigm to understand learning from human preferences. In International Conference on Artificial Intelligence and Statistics.
  4. Jinze Bai, Shuai Bai, Yunfei Chu, Zeyu Cui, Kai Dang, et al. 2023. Qwen technical report. Preprint, arXiv:2309.16609.
  5. Keqin Bao, Jizhi Zhang, Yang Zhang, Wenjie Wang, Fuli Feng, and Xiangnan He. 2023. Tallrec: An effective and efficient tuning framework to align large language model with recommendation. In RecSys. find
  6. Jizheng Chen, Kounianhua Du, Jianghao Lin, Bo Chen, Ruiming Tang, Weinan Zhang, and Yong Yu. 2024a. Elcorec: Enhance language understanding with co-propagation of numerical and categorical features for recommendation. In CIKM. find
  7. Yongjun Chen, Zhiwei Liu, Jia Li, Julian McAuley, and Caiming Xiong. 2022. Intent contrastive learning for sequential recommendation. In WWW. find
  8. Yuxin Chen, Junfei Tan, An Zhang, Zhengyi Yang, Leheng Sheng, Enzhi Zhang, Xiang Wang, and Tat-Seng Chua. 2024b. On softmax direct preference optimization for recommendation. In NeurIPS. find
  9. Xingjian Diao, Zheyuan Liu, Chunhui Zhang, Weiyi Wu, Keyi Kong, Lin Shi, Kaize Ding, Soroush Vosoughi, and Jiang Gui. 2026. Addressing overthinking in large vision-language models via gated perception-reasoning optimization. arXiv preprint arXiv:2601.04442. find
  10. Xingjian Diao, Chunhui Zhang, Keyi Kong, Weiyi Wu, Chiyu Ma, Zhongyu Ouyang, Peijun Qing, Soroush Vosoughi, and Jiang Gui. 2025. Soundmind: Rl-incentivized logic reasoning for audio-language models. In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing.
  11. Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Amy Yang, Angela Fan, et al. 2024. The llama 3 herd of models. arXiv preprint arXiv:2407.21783.
  12. Kawin Ethayarajh, Winnie Xu, Niklas Muennighoff, Dan Jurafsky, and Douwe Kiela. 2024. Kto: Model alignment as prospect theoretic optimization. arXiv preprint arXiv:2402.01306. find
  13. Xinyan Fan, Zheng Liu, Jianxun Lian, Wayne Xin Zhao, Xing Xie, and Ji-Rong Wen. 2021. Lighter and better: low-rank decomposed self-attention networks for next-item recommendation. In SIGIR. find
  14. Yunfan Gao, Tao Sheng, Youlin Xiang, Yun Xiong, Haofen Wang, and Jiawei Zhang. 2023. Chat-rec: Towards interactive and explainable llms-augmented recommender system. arXiv preprint arXiv:2303.14524. — [arXiv:2303.14524](https://arxiv.org/abs/2303.14524) find
  15. Ralph Allan Bradley and Milton E Terry. 1952. Rank analysis of incomplete block designs: I. the method of paired comparisons. Biometrika. find
  16. Shijie Geng, Shuchang Liu, Zuohui Fu, Yingqiang Ge, and Yongfeng Zhang. 2022. Recommendation as language processing (rlp): A unified pretrain, personalized prompt & predict paradigm (p5). In RecSys. find
  17. Oscar Celma. 2010. Music recommendation and discovery in the long tail. Technical report, Universitat Pompeu Fabra. find
  18. F Maxwell Harper and Joseph A Konstan. 2015. The movielens datasets: History and context. Acm transactions on interactive intelligent systems (tiis). find
  19. Jianxin Chang, Chen Gao, Yu Zheng, Yiqun Hui, Yanan Niu, Yang Song, Depeng Jin, and Yong Li. 2021. Sequential recommendation with graph neural networks. In SIGIR. find
  20. Jesse Harte, Wouter Zorgdrager, Panos Louridas, As-terios Katsifodimos, Dietmar Jannach, and Marios Fragkoulis. 2023. Leveraging large language models for sequential recommendation. In RecSys. find
  21. Ruining He and Julian McAuley. 2016. Vbpr: visual bayesian personalized ranking from implicit feedback. In AAAI. find
  22. Jianmo Ni, Jiacheng Li, and Julian McAuley. 2019. Justifying recommendations using distantly-labeled reviews and fine-grained aspects. In EMNLP-IJCNLP.
  23. Xiangnan He, Kuan Deng, Xiang Wang, Yan Li, Yongdong Zhang, and Meng Wang. 2020. Lightgcn: Simplifying and powering graph convolution network for recommendation. In ACM SIGIR conference. find
  24. Long Ouyang, Jeffrey Wu, Xu Jiang, et al. 2022. Training language models to follow instructions with human feedback. NeurIPS.
  25. B Hidasi. 2016. Session-based recommendations with recurrent neural networks. In ICLR. find
  26. Zhongyu Ouyang, Mingxuan Ju, Soroush Vosoughi, and Yanfang Ye. 2025a. Non-parametric graph convolution for re-ranking in recommendation systems. In ACM Recommender Systems conference. find
  27. Yaning Jia, Chunhui Zhang, Xingjian Diao, Xiangchi Yuan, Zhongyu Ouyang, Chiyu Ma, and Soroush Vosoughi. 2025. What makes a good curriculum? disentangling the effects of data ordering on llm mathematical reasoning. arXiv preprint arXiv:2510.19099.
  28. Zhongyu Ouyang, Chunhui Zhang, Shifu Hou, Chuxu Zhang, and Yanfang Ye. 2024. How to improve representation alignment and uniformity in graph-based collaborative filtering? In The international AAAI conference on web and social media. find
  29. Wang-Cheng Kang and Julian McAuley. 2018. Self-attentive sequential recommendation. In ICDM. find
  30. Zhongyu Ouyang, Chunhui Zhang, Yaning Jia, and Soroush Vosoughi. 2025b. Scaled supervision is an implicit lipschitz regularizer. In The International AAAI Conference on Web and Social Media. find
  31. Jure Leskovec and Julian Mcauley. 2012. Learning to discover social circles in ego networks. In NeurIPS. find
  32. Jiacheng Li, Ming Wang, Jin Li, Jinmiao Fu, Xin Shen, Jingbo Shang, and Julian McAuley. 2023. Text is all you need: Learning language representations for sequential recommendation. In SIGKDD. find
  33. Jiacheng Li, Yujie Wang, and Julian McAuley. 2020. Time interval aware self-attention for sequential recommendation. In WSDM. find
  34. Yaoyiran Li, Xiang Zhai, Moustafa Alzantot, Keyi Yu, Ivan Vulić, Anna Korhonen, and Mohamed Hammad. 2024. Calrec: Contrastive alignment of generative llms for sequential recommendation. In Proceedings of the 18th ACM Conference on Recommender Systems, pages 422–432. find
  35. Jiayi Liao, Sihang Li, Zhengyi Yang, Jiancan Wu, Yancheng Yuan, Xiang Wang, and Xiangnan He. 2024. Llara: Large language-recommendation assistant. In SIGIR. find
  36. R Duncan Luce. 1959. Individual choice behavior, volume 4. Wiley New York. find
  37. Hanjia Lyu, Song Jiang, Hanqing Zeng, Yinglong Xia, Qifan Wang, Si Zhang, Ren Chen, Christopher Leung, Jiajie Tang, and Jiebo Luo. 2023. Llm-rec: Personalized recommendation via prompting large language models. arXiv preprint arXiv:2307.15780. find
  38. Chiyu Ma, Shuo Yang, Kexin Huang, Jinda Lu, Haoming Meng, Shangshang Wang, Bolin Ding, Soroush Vosoughi, Guoyin Wang, and Jingren Zhou. 2026. Fipo: Eliciting deep reasoning with future-kl influenced policy optimization. arXiv preprint arXiv:2603.19835. — [arXiv:2603.19835](https://arxiv.org/abs/2603.19835) find
  39. Yu Meng, Mengzhou Xia, and Danqi Chen. 2024. SimPO: Simple preference optimization with a reference-free reward. In NeurIPS. find
  40. Tianyu Pang, Yujie Fang, Zihang Liu, Shenyang Deng, Lei Hsiung, Shuhua Yu, and Yaoqing Yang. 2026. Htmuon: Improving muon via heavy-tailed spectral correction. arXiv preprint arXiv:2603.10067.
  41. Robin L Plackett. 1975. The analysis of permutations. Journal of the Royal Statistical Society Series C: Applied Statistics, 24(2):193–202. find
  42. Rafael Rafailov, Archit Sharma, Eric Mitchell, Christopher D Manning, Stefano Ermon, and Chelsea Finn. 2024. Direct preference optimization: Your language model is secretly a reward model. In NeurIPS.
  43. Steffen Rendle, Christoph Freudenthaler, Zeno Gantner, and Lars Schmidt-Thieme. 2009. Bpr: Bayesian personalized ranking from implicit feedback. In The Twenty-Fifth Conference on Uncertainty in Artificial Intelligence. find
  44. John Schulman, Filip Wolski, Prafulla Dhariwal, Alec Radford, and Oleg Klimov. 2017. Proximal policy optimization algorithms. arXiv preprint arXiv:1707.06347. find
  45. Nisan Stiennon, Long Ouyang, Jeffrey Wu, Daniel Ziegler, Ryan Lowe, Chelsea Voss, Alec Radford, Dario Amodei, and Paul F Christiano. 2020. Learning to summarize with human feedback. In NeurIPS. find
  46. Jiaxi Tang and Ke Wang. 2018. Personalized top-n sequential recommendation via convolutional sequence embedding. In WSDM. find
  47. Jiancan Wu, Xiang Wang, Fuli Feng, Xiangnan He, Liang Chen, Jianxun Lian, and Xing Xie. 2021. Self-supervised graph learning for recommendation. In ACM SIGIR conference. find
  48. Xu Xie, Fei Sun, Zhaoyang Liu, Shiwen Wu, Jinyang Gao, Jiandong Zhang, Bolin Ding, and Bin Cui. 2022. Contrastive learning for sequential recommendation. In ICDE. find
  49. Haoran Xu, Amr Sharaf, Yunmo Chen, Weiting Tan, Lingfeng Shen, Benjamin Van Durme, Kenton Murray, and Young Jin Kim. 2024. Contrastive preference optimization: Pushing the boundaries of llm performance in machine translation. arXiv preprint arXiv:2401.08417.
  50. Shenghao Yang, Weizhi Ma, Peijie Sun, Qingyao Ai, Yiqun Liu, Mingchen Cai, and Min Zhang. 2024. Sequential recommendation with latent relations based on large language model. In SIGIR. find
  51. Kun Zhou, Hui Wang, Wayne Xin Zhao, Yutao Zhu, Sirui Wang, Fuzheng Zhang, Zhongyuan Wang, and Ji-Rong Wen. 2020. S3-rec: Self-supervised learning for sequential recommendation with mutual information maximization. In CIKM. find
  52. Daniel M Ziegler, Nisan Stiennon, Jeffrey Wu, Tom B Brown, Alec Radford, Dario Amodei, Paul Christiano, and Geoffrey Irving. 2019. Fine-tuning language models from human preferences. arXiv preprint arXiv:1909.08593. find
  53. Feng Yu, Yanqiao Zhu, Qiang Liu, Shu Wu, Liang Wang, and Tieniu Tan. 2020. Tagnn: Target attentive graph neural networks for session-based recommendation. In SIGIR. find
  54. Xiangchi Yuan, Xiang Chen, Tong Yu, Dachuan Shi, Can Jin, Wenke Lee, and Saayan Mitra. 2025a. Mitigating forgetting between supervised and reinforcement learning yields stronger reasoners. arXiv preprint arXiv:2510.04454. — [arXiv:2510.04454](https://arxiv.org/abs/2510.04454) find
  55. Xiangchi Yuan, Dachuan Shi, Chunhui Zhang, Zheyuan Liu, Shenglong Yao, Soroush Vosoughi, and Wenke Lee. 2026. Behavior knowledge merge in reinforced agentic models. arXiv preprint arXiv:2601.13572.
  56. Xiangchi Yuan, Chunhui Zhang, Zheyuan Liu, Dachuan Shi, Leyan Pan, Soroush Vosoughi, and Wenke Lee. 2025b. Superficial self-improved reasoners benefit from model merging. In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing.
  57. Zheng Yuan, Fajie Yuan, Yu Song, Youhua Li, Junchen Fu, Fei Yang, Yunzhu Pan, and Yongxin Ni. 2023. Where to go next for recommender systems? id-vs. modality-based recommender models revisited. In SIGIR. find
  58. Chunhui Zhang, Yiren Jian, Zhongyu Ouyang, and Soroush Vosoughi. 2025a. Pretrained image-text models are secretly video captioners. In Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies. find
  59. Chunhui Zhang, Zhongyu Ouyang, Xingjian Diao, Zheyuan Liu, and Soroush Vosoughi. 2025b. Knowing more, acting better: Hierarchical representation for embodied decision-making. In Findings of the Association for Computational Linguistics: EMNLP 2025. find
  60. Chunhui Zhang, Zhongyu Ouyang, Kwonjoon Lee, Nakul Agarwal, Sean Dae Houlihan, Soroush Vosoughi, and Shao-Yuan Lo. 2025c. Overcoming multi-step complexity in multimodal theory-of-mind reasoning: A scalable bayesian planner. arXiv preprint arXiv:2506.01301. — [arXiv:2506.01301](https://arxiv.org/abs/2506.01301) find
  61. Junjie Zhang, Ruobing Xie, Yupeng Hou, Xin Zhao, Leyu Lin, and Ji-Rong Wen. 2023. Recommendation as instruction following: A large language model empowered recommendation approach. ACM Transactions on Information Systems. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/ouyang-2026-what-makes-llms-effective-sequential/