Elamur Research⌕
☰
home / papers / 2026 / DMRetriever: A Family of Models for Improved Text Retrieval in Disaster Management

DMRetriever: A Family of Models for Improved Text Retrieval in Disaster Management

Paper · 2026 · Kai Yin, Xiangjue Dong, Chengkai Liu et al.

Summary

General-purpose embedding models perform unevenly when used for disaster management search. A model may work well for one kind of request, such as checking a claim, and fail on another, such as finding relevant tweets. DMRetriever addresses this with six dense retrievers ranging from 33M to 7.6B parameters, built on MiniLM, BERT, and Qwen3 backbones.

The training data comes from a refinement pipeline. GPT-4o-mini generates query–passage pairs from disaster-domain documents. The pairs are kept only when several strong IR models agree on them. Hard negatives are then mined at controllable difficulty levels.

Training has three stages:

  • Decoder-only backbones are converted to bidirectional attention using an MLM objective.
  • All models undergo contrastive pre-training on the unlabeled pairs.
  • Instruction fine-tuning follows, with intent-specific prompts and negatives that grow progressively harder.

Across six DisastIR search intents, every model size sets a new best score among its peers. The models are also highly parameter-efficient. For responders, this means a single domain-specific family can serve both edge devices and central command systems, and the released code, data, and checkpoints make the work reproducible.

Key points

  • The six search intents are QA, QAdoc, Twitter, FactCheck, NLI, and STS. The six models are DMRetriever-33M, -109M, -335M, -596M, -4B, and -7.6B.
  • The 596M variant outperforms XL baselines (≥4B) that are more than 13.3× its size. The 33M variant beats medium-sized baselines while using only 7.6% of their parameters.
  • Ablations on the data pipeline:
    • Top-N mutual-agreement filtering beats both no filtering and a consistency-based filter, adding up to 2.2 average points.
    • Disaster-domain training data gives up to 2.9 points on Twitter.
  • Ablations on training:
    • Bidirectional attention adds up to 2.7 points for the 596M model.
    • Intent instructions add up to 2.3 points for the 109M model.
    • Skipping the progressive curriculum hurts small models.
  • The work is limited to English text. Multimodal and multilingual retrieval are left for future work.

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

🎓Texas A&M UniversityCollege Station · United States26 works · 1,871 citations

 Fields, topics and concepts

📚 Shelves 7The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 worksSocial Sciences · 1,772 worksInformation Systems · 1,437 worksSociology and Political Science · 577 works

🧭 Topics 1Research subjects and languages the work is about.Information retrieval & search · 473 works

Cite

Cite as Kai Yin, Xiangjue Dong, Chengkai Liu, Allen Lin, Lingfeng Shi, Ali Mostafavi and James Caverlee. 2026. DMRetriever: A Family of Models for Improved Text Retrieval in Disaster Management. 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.722

@inproceedings{yin2026dmretriever,
  title = {{DMRetriever: A Family of Models for Improved Text Retrieval in Disaster Management}},
  author = {Kai Yin, Xiangjue Dong, Chengkai Liu, Allen Lin, Lingfeng Shi, Ali Mostafavi and James Caverlee},
  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.722},
  url = {https://doi.org/10.18653/v1/2026.acl-long.722}
}

References

55 references, 29 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Reem Abbas and Todd Miller. 2025. Exploring communication inefficiencies in disaster response: Perspectives of emergency managers and health professionals. International Journal of Disaster Risk Reduction, 120:105393. find
  2. Firoj Alam, Hassan Sajjad, Muhammad Imran, and Ferda Ofli. 2021. Crisisbench: Benchmarking crisis-related social media datasets for humanitarian information processing. In Proceedings of the International AAAI conference on web and social media, volume 15, pages 923–932. find
  3. Chris Alberti, Daniel Andor, Emily Pitler, Jacob Devlin, and Michael Collins. 2019. Synthetic qa corpora generation with roundtrip consistency. arXiv preprint arXiv:1906.05416.
  4. Akari Asai, Timo Schick, Patrick Lewis, Xilun Chen, Gautier Izacard, Sebastian Riedel, Hannaneh Hajishirzi, and Wen-tau Yih. 2022. Task-aware retrieval with instructions. arXiv preprint arXiv:2211.09260.
  5. Parul Awasthy, Aashka Trivedi, Yulong Li, Mihaela Bornea, David Cox, Abraham Daniels, Martin Franz, Gabe Goodhart, Bhavani Iyer, Vishwajeet Kumar, and 1 others. 2025. Granite embedding models. arXiv preprint arXiv:2502.20204. — [arXiv:2502.20204](https://arxiv.org/abs/2502.20204) find
  6. Parishad BehnamGhader, Vaibhav Adlakha, Marius Mosbach, Dzmitry Bahdanau, Nicolas Chapados, and Siva Reddy. 2024. Llm2vec: Large language models are secretly powerful text encoders. arXiv preprint arXiv:2404.05961.
  7. Yoshua Bengio, Jérôme Louradour, Ronan Collobert, and Jason Weston. 2009. Curriculum learning. In Proceedings of the 26th annual international conference on machine learning, pages 41–48.
  8. Vladimir Karpukhin, Barlas Oguz, Sewon Min, Patrick SH Lewis, Ledell Wu, Sergey Edunov, Danqi Chen, and Wen-tau Yih. 2020. Dense passage retrieval for open-domain question answering. In EMNLP (1), pages 6769–6781.
  9. Junseong Kim, Seolhwa Lee, Jihoon Kwon, Sangmo Gu, Yejin Kim, Minkyung Cho, Jy yong Sohn, and Chanyeol Choi. 2024. Linq-embed-mistral:elevating text retrieval with improved gpt data through task-specific control and quality refinement. Linq AI Research Blog. find
  10. Vera Boteva, Demian Gholipour, Artem Sokolov, and Stefan Riezler. 2016. A full-text learning to rank dataset for medical information retrieval. In European Conference on Information Retrieval, pages 716–722. Springer. find
  11. Chankyu Lee, Rajarshi Roy, Mengyao Xu, Jonathan Raiman, Mohammad Shoeybi, Bryan Catanzaro, and Wei Ping. 2024a. Nv-embed: Improved techniques for training llms as generalist embedding models. arXiv preprint arXiv:2405.17428.
  12. Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, and 1 others. 2020. Language models are few-shot learners. Advances in neural information processing systems, 33:1877–1901.
  13. Sean Lee, Aamir Shakir, Darius Koenig, and Julius Lipp. 2024b. Open source strikes bread - new fluffy embeddings model. find
  14. Ilias Chalkidis, Manos Fergadiotis, Prodromos Malakasiotis, Nikolaos Aletras, and Ion Androutsopoulos. 2020. Legal-bert: The muppets straight out of law school. arXiv preprint arXiv:2010.02559.
  15. Ting Chen, Simon Kornblith, Mohammad Norouzi, and Geoffrey Hinton. 2020. A simple framework for contrastive learning of visual representations. In International conference on machine learning, pages 1597–1607. PmLR.
  16. Zhitong Chen, Kai Yin, Xiangjue Dong, Chengkai Liu, Xiangpeng Li, Yiming Xiao, Bo Li, Junwei Ma, Ali Mostafavi, and James Caverlee. 2026. Disastqa: A comprehensive benchmark for evaluating question answering in disaster management. arXiv preprint arXiv:2601.03670. find
  17. Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. Bert: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of the 2019 conference of the North American chapter of the association for computational linguistics: human language technologies,…
  18. Bo Li, Junwei Ma, Kai Yin, Yiming Xiao, Chia-Wei Hsu, and Ali Mostafavi. 2025. Disaster management in the era of agentic ai systems: A vision for collective human-machine intelligence for augmented resilience. arXiv preprint arXiv:2510.16034. — [arXiv:2510.16034](https://arxiv.org/abs/2510.16034) find
  19. Zehan Li, Xin Zhang, Yanzhao Zhang, Dingkun Long, Pengjun Xie, and Meishan Zhang. 2023. Towards general text embeddings with multi-stage contrastive learning. Preprint, arXiv:2308.03281.
  20. Hanzhou Liu, Kai Yin, Zhitong Chen, Chenyue Liu, and Ali Mostafavi. 2025a. Floodsql-bench: A retrieval-augmented benchmark for geospatially-grounded text-to-sql. arXiv preprint arXiv:2512.12084. — [arXiv:2512.12084](https://arxiv.org/abs/2512.12084) find
  21. Zhewei Liu, Natalie Coleman, Flavia Ioana Patrascu, Kai Yin, Xiangpeng Li, and Ali Mostafavi. 2025b. Artificial intelligence for flood risk management: A comprehensive state-of-the-art review and future directions. International Journal of Disaster Risk Reduction, 117:105110. find
  22. Rui Meng, Ye Liu, Shafiq Rayhan Joty, Caiming Xiong, Yingbo Zhou, and Semih Yavuz. 2024. Sfr-embedding-mistral:enhance text retrieval with transfer learning. Salesforce AI Research Blog. find
  23. Tianyu Gao, Xingcheng Yao, and Danqi Chen. 2021. Simcse: Simple contrastive learning of sentence embeddings. arXiv preprint arXiv:2104.08821.
  24. Stephen Merity, Caiming Xiong, James Bradbury, and Richard Socher. 2016. Pointer sentinel mixture models. Preprint, arXiv:1609.07843.
  25. Lipai Huang, Kai Yin, Chia-Fu Liu, and Ali Mostafavi. 2026. R2rag-flood: A reasoning-reinforced training-free retrieval augmentation generation framework for flood damage nowcasting. arXiv preprint arXiv:2602.10312. find
  26. Luke Merrick, Danmei Xu, Gaurav Nuti, and Daniel Campos. 2024. Arctic-embed: Scalable, efficient, and accurate text embedding models. arXiv preprint arXiv:2405.05374. — [arXiv:2405.05374](https://arxiv.org/abs/2405.05374) find
  27. Muhammad Imran, Carlos Castillo, Fernando Diaz, and Sarah Vieweg. 2015. Processing social media messages in mass emergency: A survey. ACM computing surveys (CSUR), 47(4):1–38. find
  28. Gabriel de Souza P Moreira, Radek Osmulski, Mengyao Xu, Ronay Ak, Benedikt Schifferer, and Even Oldridge. 2024. Nv-retriever: Improving text embedding models with effective hard-negative mining. arXiv preprint arXiv:2407.15831. — [arXiv:2407.15831](https://arxiv.org/abs/2407.15831) find
  29. Niklas Muennighoff, SU Hongjin, Liang Wang, Nan Yang, Furu Wei, Tao Yu, Amanpreet Singh, and Douwe Kiela. 2024. Generative representational instruction tuning. In The Thirteenth International Conference on Learning Representations.
  30. UNDRR. 2020. Hazard definition and classification review technical report. Technical report, United Nations Office for Disaster Risk Reduction, Geneva, Switzerland. Supported by BMZ and USAID. Chair: Professor Virginia Murray. find
  31. Niklas Muennighoff, Nouamane Tazi, Loïc Magne, and Nils Reimers. 2022. Mteb: Massive text embedding benchmark. arXiv preprint arXiv:2210.07316.
  32. Liang Wang, Nan Yang, Xiaolong Huang, Binxing Jiao, Linjun Yang, Daxin Jiang, Rangan Majumder, and Furu Wei. 2022. Text embeddings by weakly-supervised contrastive pre-training. arXiv preprint arXiv:2212.03533.
  33. Tri Nguyen, Mir Rosenberg, Xia Song, Jianfeng Gao, Saurabh Tiwary, Rangan Majumder, and Li Deng. 2016. Ms marco: A human-generated machine reading comprehension dataset.
  34. 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, 35:27730–27744.
  35. Shalini Priya, Manish Bhanu, Sourav Kumar Dandapat, Kripabandhu Ghosh, and Joydeep Chandra. 2020. Taqe: tweet retrieval-based infrastructure damage assessment during disasters. IEEE transactions on computational social systems, 7(2):389–403. find
  36. Colin Raffel, Noam Shazeer, Adam Roberts, Katherine 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 research, 21(140):1–67.
  37. Jeff Rasley, Samyam Rajbhandari, Olatunji Ruwase, and Yuxiong He. 2020. Deepspeed: System optimizations enable training deep learning models with over 100 billion parameters. In Proceedings of the 26th ACM SIGKDD international conference on knowledge discovery & data mining, pages 3505–3506. find
  38. Nils Reimers and Iryna Gurevych. 2019. Sentence-bert: Sentence embeddings using siamese bert-networks. arXiv preprint arXiv:1908.10084.
  39. Oscar Sainz, Jon Ander Campos, Iker García-Ferrero, Julen Etxaniz, Oier Lopez de Lacalle, and Eneko Agirre. 2023. Nlp evaluation in trouble: On the need to measure llm data contamination for each benchmark. arXiv preprint arXiv:2310.18018.
  40. Bhaskarjit Sarmah, Dhagash Mehta, Benika Hall, Rohan Rao, Sunil Patel, and Stefano Pasquali. 2024. Hybridrag: Integrating knowledge graphs and vector retrieval augmented generation for efficient information extraction. In Proceedings of the 5th ACM International Conference on AI in Finance, pages 60… find
  41. Nandan Thakur, Nils Reimers, Andreas Rücklé, Abhishek Srivastava, and Iryna Gurevych. 2021. Beir: A heterogenous benchmark for zero-shot evaluation of information retrieval models. arXiv preprint arXiv:2104.08663.
  42. Liang Wang, Nan Yang, Xiaolong Huang, Linjun Yang, Rangan Majumder, and Furu Wei. 2023. Improving text embeddings with large language models. arXiv preprint arXiv:2401.00368.
  43. Liang Wang, Nan Yang, Xiaolong Huang, Linjun Yang, Rangan Majumder, and Furu Wei. 2024. Multilingual e5 text embeddings: A technical report. arXiv preprint arXiv:2402.05672.
  44. Jason Wei, Maarten Bosma, Vincent Y Zhao, Kelvin Guu, Adams Wei Yu, Brian Lester, Nan Du, Andrew M Dai, and Quoc V Le. 2021. Finetuned language models are zero-shot learners. arXiv preprint arXiv:2109.01652.
  45. Shitao Xiao, Zheng Liu, Peitian Zhang, and Niklas Muennighoff. 2023. C-pack: Packaged resources to advance general chinese embedding. Preprint, arXiv:2309.07597.
  46. Shitao Xiao, Zheng Liu, Peitian Zhang, Niklas Muennighoff, Defu Lian, and Jian-Yun Nie. 2024. C-pack: Packed resources for general chinese embeddings. In Proceedings of the 47th international ACM SIGIR conference on research and development in information retrieval, pages 641–649.
  47. Yiming Xiao, Kai Yin, and Ali Mostafavi. 2026. Crisisense-rag: Crisis sensing multimodal retrieval-augmented generation for rapid disaster impact assessment. arXiv preprint arXiv:2602.13239. — [arXiv:2602.13239](https://arxiv.org/abs/2602.13239) find
  48. Ran Xu, Hui Liu, Sreyashi Nag, Zhenwei Dai, Yaochen Xie, Xianfeng Tang, Chen Luo, Yang Li, Joyce C Ho, Carl Yang, and 1 others. 2024a. Simrag: Self-improving retrieval-augmented generation for adapting large language models to specialized domains. arXiv preprint arXiv:2410.17952. find
  49. Ran Xu, Wenqi Shi, Yue Yu, Yuchen Zhuang, Yanqiao Zhu, May D Wang, Joyce C Ho, Chao Zhang, and Carl Yang. 2024b. Bmretriever: Tuning large language models as better biomedical text retrievers. arXiv preprint arXiv:2404.18443. find
  50. Junhan Yang, Jiahe Wan, Yichen Yao, Wei Chu, Yinghui Xu, and Yuan Qi. 2025. inf-retriever-v1 (revision 5f469d7). find
  51. Zhilin Yang, Peng Qi, Saizheng Zhang, Yoshua Bengio, William W Cohen, Ruslan Salakhutdinov, and Christopher D Manning. 2018. Hotpotqa: A dataset for diverse, explainable multi-hop question answering. arXiv preprint arXiv:1809.09600.
  52. Kai Yin, Xiangjue Dong, Chengkai Liu, Lipai Huang, Yiming Xiao, Zhewei Liu, Ali Mostafavi, and James Caverlee. 2025. Disastir: A comprehensive information retrieval benchmark for disaster management. arXiv preprint arXiv:2505.15856. find
  53. Kai Yin, Bo Li, Chengkai Liu, Ali Mostafavi, and Xia Hu. 2024. Crisissense-llm: Instruction fine-tuned large language model for multi-label social media text classification in disaster informatics. arXiv preprint arXiv:2406.15477. — [arXiv:2406.15477](https://arxiv.org/abs/2406.15477) find
  54. Yanzhao Zhang, Mingxin Li, Dingkun Long, Xin Zhang, Huan Lin, Baosong Yang, Pengjun Xie, An Yang, Dayiheng Liu, Junyang Lin, Fei Huang, and Jingren Zhou. 2025. Qwen3 embedding: Advancing text embedding and reranking through foundation models. arXiv preprint arXiv:2506.05176.
  55. Algorithm 1: Three-stage training pipeline of DMR ETRIEVER Input: Backbone M (encoder- or decoder-only); MLM training data D M LM ; Difficulty level schedule {α t } for progressive fine-tuning Output: Final DMR ETRIEVER model 1 Stage 1: Bidirectional attention adaptation (decoder-only backbone) 2 if… find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/yin-2026-dmretriever-family-models-improved-text/