Elamur Research⌕
☰
home / papers / 2026 / DVMap: Fine-Grained Pluralistic Value Alignment via High-Consensus Demographic-Value Mapping

DVMap: Fine-Grained Pluralistic Value Alignment via High-Consensus Demographic-Value Mapping

Paper · 2026 · Pengyun Zhu, Yuqi Ren, Zhen Wang et al.

Summary

Most pluralistic value alignment for LLMs conditions models on a country label. This ignores how much values differ inside a single nation. DVMap addresses this by conditioning on a combination of demographic attributes instead.

The authors used World Values Survey (WVS) Wave 7 responses. They kept only demographic profiles whose members answered a given question identically, using zero Shannon entropy as the consensus test. This produced a training corpus of 56,152 samples covering 10 countries and 16 value topics.

Models are then trained with Group Relative Policy Optimization (GRPO), using a simple binary correctness reward. A structured Chain-of-Thought (CoT) template makes the model reason step by step about how each attribute relates to each answer option.

The authors also built a 21,553-sample benchmark that tests generalization to unseen demographics, countries and values. A Qwen3-8B model trained this way outperforms DeepSeek-v3.2 on the unseen-demographics test.

The work shows that intermediate-granularity, consensus-filtered supervision gives LLMs a better grasp of group-level value differences than coarse national prompting does.

Key points

  • Data: the corpus comes from WVS Wave 7 and uses 11-dimensional demographic profiles (e.g., age, income, religion, language). The countries were chosen with the help of the Inglehart-Welzel Cultural Map.
  • Headline result: on the cross-demographic test, Qwen3-8B-DVMap reaches 48.6% accuracy, versus 45.1% for DeepSeek-v3.2.
  • Filtering ablation (Qwen3-4B):
    • Strict consensus filtering (H = 0) reaches 47.9% accuracy and a Wasserstein Distance (WD) of 0.142.
    • Looser majority-style filtering (H ≥ 0) reaches 46.5% accuracy and a WD of 0.149.
    • The base model scores 44.3% accuracy and a WD of 0.158.
  • Reasoning and reward ablations (Qwen3-4B):
    • Using CoT only at inference time lowers accuracy to 43.5%.
    • RL with free-form reasoning reaches 46.2%.
    • A Likert-scaled soft reward reaches 46.3%, with a WD of 0.155.
    • All three trail DVMap.
  • Limitations and release: the stated limitations are that WVS is static, some groups may be underrepresented, and evaluation uses only multiple-choice questions. Code and data are available at github.com/EnlightenedAI/DVMap.

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

🎓Tianjin UniversityTianjin · China52 works · 1,128 citations

 Fields, topics and concepts

📚 Shelves 8The field of science this work belongs to (OpenAlex fields).Computer Science · 18,614 worksArtificial Intelligence · 17,965 worksArts and Humanities · 17,450 worksLanguage and Linguistics · 17,296 worksSocial Sciences · 1,954 worksPsychology · 1,649 worksSafety Research · 626 worksSocial Psychology · 330 works

Cite

Cite as Pengyun Zhu, Yuqi Ren, Zhen Wang, Lei Yang and Deyi Xiong. 2026. DVMap: Fine-Grained Pluralistic Value Alignment via High-Consensus Demographic-Value Mapping. 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.909

@inproceedings{zhu2026dvmap,
  title = {{DVMap: Fine-Grained Pluralistic Value Alignment via High-Consensus Demographic-Value Mapping}},
  author = {Pengyun Zhu, Yuqi Ren, Zhen Wang, Lei Yang and Deyi Xiong},
  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.909},
  url = {https://doi.org/10.18653/v1/2026.acl-long.909}
}

References

50 references, 21 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Badr AlKhamissi, Muhammad ElNokrashy, Mai Alkhamissi, and Mona Diab. 2024. Investigating cultural alignment of large language models. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 12404–12422.
  2. Jacob Andreas. 2022. Language models as agent models. In Findings of the Association for Computational Linguistics: EMNLP 2022, pages 5769–5779.
  3. Anthropic. 2025. Claude 3.7 sonnet. find
  4. Sanjeev Arora and Anirudh Goyal. 2023. A theory for emergence of complex skills in language models. CoRR, abs/2307.15936. — [arXiv:2307.15936](https://arxiv.org/abs/2307.15936) find
  5. Amanda Askell, Yuntao Bai, Anna Chen, Dawn Drain, Deep Ganguli, Tom Henighan, Andy Jones, Nicholas Joseph, Benjamin Mann, Nova DasSarma, Nelson Elhage, Zac Hatfield-Dodds, Danny Hernandez, Jackson Kernion, Kamal Ndousse, Catherine Olsson, Dario Amodei, Tom B. Brown, Jack Clark, and 3 others. 2021. A… find
  6. Yuntao Bai, Andy Jones, Kamal Ndousse, Amanda Askell, Anna Chen, Nova DasSarma, Dawn Drain, Stanislav Fort, Deep Ganguli, Tom Henighan, Nicholas Joseph, Saurav Kadavath, Jackson Kernion, Tom Conerly, Sheer El Showk, Nelson Elhage, Zac Hatfield-Dodds, Danny Hernandez, Tristan Hume, and 12 others. 202… find
  7. Pierre Bourdieu. 2018. The forms of capital. In The sociology of economic life, pages 78–92. Routledge. find
  8. Leo Breiman. 2001. Random forests. Machine learning, 45(1):5–32. find
  9. Yong Cao, Li Zhou, Seolhwa Lee, Laura Cabello Piqueras, Min Chen, and Daniel Hershcovich. 2023. Assessing cross-cultural alignment between chatgpt and human societies: An empirical study. In Proceedings of the first workshop on cross-cultural considerations in NLP (C3NLP), pages 53–67.
  10. Rochelle Choenni, Anne Lauscher, and Ekaterina Shutova. 2024. The echoes of multilinguality: Tracing cultural value shifts during language model fine-tuning. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 15042–15058. find
  11. Rochelle Choenni and Ekaterina Shutova. 2024. Self-alignment: Improving alignment of cultural values in LLMs via in-context learning. CoRR, abs/2408.16482. — [arXiv:2408.16482](https://arxiv.org/abs/2408.16482) find
  12. DeepSeek-AI. 2024. Deepseek-v3 technical report. CoRR, abs/2412.19437. find
  13. Esin Durmus, Karina Nyugen, Thomas I. Liao, Nicholas Schiefer, Amanda Askell, Anton Bakhtin, Carol Chen, Zac Hatfield-Dodds, Danny Hernandez, Nicholas Joseph, Liane Lovitt, Sam McCandlish, Orowa Sikder, Alex Tamkin, Janel Thamkul, Jared Kaplan, Jack Clark, and Deep Ganguli. 2023. Towards measuring t…
  14. Shangbin Feng, Taylor Sorensen, Yuhan Liu, Jillian Fisher, Chan Young Park, Yejin Choi, and Yulia Tsvetkov. 2024. Modular pluralism: Pluralistic alignment via multi-llm collaboration. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 4151–4171.
  15. Leo Gao, Stella Biderman, Sid Black, Laurence Golding, Travis Hoppe, Charles Foster, Jason Phang, Horace He, Anish Thite, Noa Nabeshima, Shawn Presser, and Connor Leahy. 2021. The pile: An 800gb dataset of diverse text for language modeling. CoRR, abs/2101.00027. find
  16. Aaron Grattafiori, Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Alex Vaughan, and 1 others. 2024. The llama 3 herd of models. In Neural Information Processing Systems. Curran Associates.
  17. Jian Guan, Junfei Wu, Jia-Nan Li, Chuanqi Cheng, and Wei Wu. 2025. A survey on personalized alignment—the missing piece for large language models in real-world applications. In Findings of the Association for Computational Linguistics: ACL 2025, pages 5313–5333. find
  18. C. Haerpfer, R. Inglehart, A. Moreno, C. Welzel, K. Kizilova, J. Diez-Medrano, M. Lagos, P. Norris, E. Ponarin, and B. Puranen. 2022. World values survey: Round seven – country-pooled datafile version 6.0. JD Systems Institute & WVSA Secretariat, Madrid, Spain & Vienna, Austria. find
  19. Zihao He, Siyi Guo, Ashwin Rao, and Kristina Lerman. 2024. Whose emotions and moral sentiments do language models reflect? In Findings of the Association for Computational Linguistics, ACL 2024, Bangkok, Thailand and virtual meeting, August 11-16, 2024, Findings of ACL, pages 6611–6631. Association… find
  20. Dan Hendrycks, Collin Burns, Steven Basart, Andrew Critch, Jerry Li, Dawn Song, and Jacob Steinhardt. 2021. Aligning AI with shared human values. In 9th International Conference on Learning Representations, ICLR 2021, Virtual Event, Austria, May 3-7, 2021. OpenReview.net.
  21. Ronald Inglehart and Christian Welzel. 2005. Modernization, cultural change, and democracy. The human development sequence. find
  22. Rebecca L. Johnson, Giada Pistilli, Natalia Menédez-González, Leslye Denisse Dias Duran, Enrico Panai, Julija Kalpokiene, and Donald Jay Bertulfo. 2022. The ghost in the machine has an american accent: value conflict in GPT-3. CoRR, abs/2203.07785.
  23. Enkelejda Kasneci, Kathrin Seßler, Stefan Küchemann, Maria Bannert, Daryna Dementieva, Frank Fischer, Urs Gasser, Georg Groh, Stephan Günnemann, Eyke Hüllermeier, and 1 others. 2023. Chatgpt for good? on opportunities and challenges of large language models for education. Learning and individual dif…
  24. Daniel Keysers, Nathanael Schärli, Nathan Scales, Hylke Buisman, Daniel Furrer, Sergii Kashubin, Nikola Momchev, Danila Sinopalnikov, Lukasz Stafiniak, Tibor Tihon, Dmitry Tsarkov, Xiao Wang, Marc van Zee, and Olivier Bousquet. 2020. Measuring compositional generalization: A comprehensive method on… find
  25. Grgur Kovac, Masataka Sawayama, Rémy Portelas, Cédric Colas, Peter Ford Dominey, and Pierre-Yves Oudeyer. 2023. Large language models as superpositions of cultural perspectives. CoRR, abs/2307.07870. — [arXiv:2307.07870](https://arxiv.org/abs/2307.07870) find
  26. Preethi Lahoti, Nicholas Blumm, Xiao Ma, Raghavendra Kotikalapudi, Sahitya Potluri, Qijun Tan, Hansa Srinivasan, Ben Packer, Ahmad Beirami, Alex Beutel, and Jilin Chen. 2023. Improving diversity of demographic representation in large language models via collective-critiques and self-voting. In Proce… find
  27. Cheng Li, Mengzhuo Chen, Jindong Wang, Sunayana Sitaram, and Xing Xie. 2024a. Culturellm: Incorporating cultural differences into large language models. Advances in Neural Information Processing Systems, 37:84799–84838.
  28. Cheng Li, Damien Teney, Linyi Yang, Qingsong Wen, Xing Xie, and Jindong Wang. 2024b. Culturepark: Boosting cross-cultural understanding in large language models. Advances in Neural Information Processing Systems, 37:65183–65216.
  29. Chen Cecilia Liu, Fajri Koto, Timothy Baldwin, and Iryna Gurevych. 2024. Are multilingual LLMs culturally-diverse reasoners? an investigation into multicultural proverbs and sayings. In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics:…
  30. Paweł Niszczota, Mateusz Janczak, and Michał Misiak. 2025. Large language models can replicate cross-cultural differences in personality. Journal of Research in Personality, 115:104584. find
  31. OpenAI. 2024. Gpt-4o system card. abs/2410.21276.
  32. CoRR, Long Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida, Carroll Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, and 1 others. 2022. Training language models to follow instructions with human feedback. Advances in neural information processing systems, 35:27730–27…
  33. Joon Sung Park, Joseph O’Brien, Carrie Jun Cai, Meredith Ringel Morris, Percy Liang, and Michael S Bernstein. 2023. Generative agents: Interactive simulacra of human behavior. In Proceedings of the 36th annual acm symposium on user interface software and technology, pages 1–22. find
  34. Jianxiang Peng, Ling Shi, Xinwei Wu, Hanwen Zhang, Fujiang Liu, Haocheng Lyu, and Deyi Xiong. 2025. DiplomacyAgent: Do LLMs balance interests and ethical principles in international events? In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing, pages 13721–13739,… find
  35. Salvatore Flavio Pileggi. 2024. A hybrid approach to analysing large scale surveys: individual values, opinions and perceptions. SN Social Sciences, 4(8):144. find
  36. Rafael Rafailov, Archit Sharma, Eric Mitchell, Christopher D Manning, Stefano Ermon, and Chelsea Finn. 2023. Direct preference optimization: Your language model is secretly a reward model. Advances in neural information processing systems, 36:53728–53741.
  37. Shibani Santurkar, Esin Durmus, Faisal Ladhak, Cinoo Lee, Percy Liang, and Tatsunori Hashimoto. 2023. Whose opinions do language models reflect? In International Conference on Machine Learning, pages 29971–30004. PMLR.
  38. Zhihong Shao, Peiyi Wang, Qihao Zhu, Runxin Xu, Junxiao Song, Mingchuan Zhang, Y. K. Li, Y. Wu, and Daya Guo. 2024. Deepseekmath: Pushing the limits of mathematical reasoning in open language models. CoRR, abs/2402.03300.
  39. Siqi Shen, Lajanugen Logeswaran, Moontae Lee, Honglak Lee, Soujanya Poria, and Rada Mihalcea. 2024. Understanding the capabilities and limitations of large language models for cultural commonsense. In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computation… find
  40. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 6349–6384. find
  41. Walter F Wiggins and Ali S Tejani. 2022. On the opportunities and risks of foundation models for natural language processing in radiology. Radiology: Artificial Intelligence, 4(4):e220119. find
  42. Shaoyang Xu, Weilong Dong, Zishan Guo, Xinwei Wu, and Deyi Xiong. 2024. Exploring multilingual concepts of human values in large language models: Is value alignment consistent, transferable and controllable across languages? In Findings of the Association for Computational Linguistics: EMNLP 2024, p… find
  43. Shaoyang Xu, Yongqi Leng, Linhao Yu, and Deyi Xiong. 2025. Self-pluralising culture alignment for large language models. In Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers),… find
  44. An Yang, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chengyuan Li, Dayiheng Liu, Fei Huang, Haoran Wei, Huan Lin, Jian Yang, Jianhong Tu, Jianwei Zhang, Jianxin Yang, Jiaxi Yang, Jingren Zhou, Junyang Lin, Kai Dang, and 22 others. 2024. Qwen2.5 technical report. CoRR, abs/2412.1511… find
  45. Tianhao Shen, Renren Jin, Yufei Huang, Chuang Liu, Weilong Dong, Zishan Guo, Xinwei Wu, Yan Liu, and Deyi Xiong. 2023a. Large language model alignment: A survey. CoRR, abs/2309.15025.
  46. Tianhao Shen, Sun Li, Quan Tu, and Deyi Xiong. 2023b. Roleeval: A bilingual role evaluation benchmark for large language models. arXiv preprint arXiv:2312.16132. — [arXiv:2312.16132](https://arxiv.org/abs/2312.16132) find
  47. Gabriel Simmons. 2023. Moral mimicry: Large language models produce moral rationalizations tailored to political identity. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 4: Student Research Workshop), pages 282–297. find
  48. Gemini Team. 2025a. Gemini 2.5: Pushing the frontier with advanced reasoning, multimodality, long context, and next generation agentic capabilities. CoRR, abs/2507.06261.
  49. Qwen Team. 2025b. Qwen3 technical report. CoRR, abs/2505.09388.
  50. Wenxuan Wang, Wenxiang Jiao, Jingyuan Huang, Ruyi Dai, Jen-tse Huang, Zhaopeng Tu, and Michael Lyu. 2024. Not all countries celebrate thanksgiving: On the cultural dominance in large language models. In

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/zhu-2026-dvmap-fine-grained-pluralistic-value/