Elamur Research⌕
☰
home / papers / 2026 / TravelBehaviorQA: A Benchmark Dataset for Behavioral Interpretation of GPS Trajectories

TravelBehaviorQA: A Benchmark Dataset for Behavioral Interpretation of GPS Trajectories

Paper · 2026 · Dongyang Zhen, Niping Duan, Huan Zhou et al.

Summary

GPS traces implicitly record how people travel, schedule activities, and form routines. However, existing mobility resources are mostly built around narrow objectives, like guessing where someone will go next or labeling which transport mode they used. TravelBehaviorQA asks whether LLMs can instead read trajectories as behavioral evidence and explain them in natural language. The authors build the benchmark on the GeoLife GPS data and pair it with about 143k question–answer instances. Trajectories are given to models as text: serialized coordinates, optionally enriched with trip segmentation, POI-based place semantics, and travel modes. The questions combine deterministic, fact-based items (counts, times, places, modes) with open-ended items that require daily, weekly, or routine-level narratives. Evaluating a range of current LLMs reveals a clear gap between extracting facts and actually interpreting behavior. Larger models do not close this gap. How the trajectory is represented matters much more than model size. The resource is public and offers a testbed for grounded, faithful reasoning over long, non-textual temporal signals.

Key points

  • Scale and coverage: 143,238 QA pairs from GeoLife, covering 68 users and 6,305 user-days (2007–2011). The questions span 11 behavioral categories and 30 question types, with 75.1% deterministic and 24.9% open-ended.
  • Comparison with prior datasets: TravelBehaviorQA is far larger than MobQA (~5.8K) and POI-QA (~10K). It also targets narrative summarization and faithfulness, not just localized fact lookup.
  • Effect of input representation: For GPT-5.2, deterministic accuracy is only 0.05 with raw GPS. It rises to 0.58 with segmented trajectories plus POI and to 0.61 with trip-level narratives plus POI.
  • Ablation: Removing segmentation hurts most, cutting deterministic accuracy from 0.35 to 0.10. Removing mode labels or POI data causes smaller drops, concentrated in the related question categories.
  • Decoding: Temperature 0.0 gives higher accuracy and better run-to-run reproducibility than temperature 1.0 on deterministic questions. The dataset is available on GitHub (dongyangzhen/TravelBehaviorQA).

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 Maryland, College ParkCollege Park · United States140 works · 29,011 citations

 Fields, topics and concepts

📚 Shelves 6The field of science this work belongs to (OpenAlex fields).Computer Science · 18,614 worksSocial Sciences · 1,954 worksEngineering · 188 worksSignal Processing · 493 worksBuilding and Construction · 7 worksTransportation · 7 works

🧭 Topics 1Research subjects and languages the work is about.Evaluation & benchmarks · 1,341 works

Cite

Cite as Dongyang Zhen, Niping Duan, Huan Zhou and Qingbin Cui. 2026. TravelBehaviorQA: A Benchmark Dataset for Behavioral Interpretation of GPS Trajectories. 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.1604

@inproceedings{zhen2026travelbehaviorqa,
  title = {{TravelBehaviorQA: A Benchmark Dataset for Behavioral Interpretation of GPS Trajectories}},
  author = {Dongyang Zhen, Niping Duan, Huan Zhou and Qingbin Cui},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  year = {2026},
  doi = {10.18653/v1/2026.findings-acl.1604},
  url = {https://doi.org/10.18653/v1/2026.findings-acl.1604}
}

References

40 references, 2 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Timur Abbiasov, Cate Heine, Sadegh Sabouri, Arianna Salazar-Miranda, Paolo Santi, Edward Glaeser, and Carlo Ratti. 2024. The 15-minute city quantified using human mobility data. Nature Human Behaviour, 8(3):445–455. find
  2. Hikaru Asano, Hiroki Ouchi, Akira Kasuga, and Ryo Yonetani. 2025. Mobqa: A benchmark dataset for semantic understanding of human mobility data through question answering. arXiv preprint arXiv:2508.11163. — [arXiv:2508.11163](https://arxiv.org/abs/2508.11163) find
  3. Derya Birant and Alp Kut. 2007. St-dbscan: An algorithm for clustering spatial–temporal data. Data & knowledge engineering, 60(1):208–221. find
  4. Yong Chen, Ben Chi, Chuanjia Li, Yuliang Zhang, Chenlei Liao, Xiqun Chen, and Na Xie. 2025. Toward interactive next location prediction driven by large language models. IEEE Transactions on Computational Social Systems. find
  5. Jiawei Cheng, Jingyuan Wang, Yichuan Zhang, Jiahao Ji, Yuanshao Zhu, Zhibo Zhang, and Xiangyu Zhao. 2025. Poi-enhancer: An llm-based semantic enhancement framework for poi representation learning. Proceedings of the AAAI Conference on Artificial Intelligence, 39(11):11509–11517. find
  6. Xiao Fu, Yi Zhang, Juan de Dios Ortúzar, and Guonian Lü. 2025. Activity-travel pattern inference based on multi-source big data. Transport Reviews, 45(1):26– 48. find
  7. Shichao Ge, Peijun Ye, Renrui Zhang, Min Zhou, Hairong Dong, and Fei-Yue Wang. 2025. Llm-driven cognitive modeling for personalized travel generation. IEEE Transactions on Computational Social Systems. find
  8. Shreya Ghosh, Soumya K Ghosh, Sajal K Das, and Prasenjit Mitra. 2024. Mobilytics: Mobility analytics framework for transferring semantic knowledge. IEEE Transactions on Mobile Computing, 23(12):11588–11603. find
  9. Letian Gong, Yan Lin, Xinyue Zhang, Yiwen Lu, Xuedi Han, Yichen Liu, Shengnan Guo, Youfang Lin, and Huaiyu Wan. 2024. Mobility-llm: Learning visiting intentions and travel preference from human mobility data with large language models. Advances in Neural Information Processing Systems, 37:36185–3621… find
  10. Xiao Han, Dayan Pan, Xiangyu Zhao, Xuyuan Hu, Zhaolin Deng, Xiangjie Kong, and Guojiang Shen. 2025. A dataset for spatiotemporal-sensitive poi question answering. arXiv preprint arXiv:2505.10928. — [arXiv:2505.10928](https://arxiv.org/abs/2505.10928) find
  11. Yu-Chung Hsiao, Fedir Zubach, Gilles Baechler, Srinivas Sunkara, Victor Cărbune, Jason Lin, Maria Wang, Yun Zhu, and Jindong Chen. 2025. Screenqa: Large-scale question-answer pairs over mobile app screenshots. In Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Assoc… find
  12. Tianhao Huang, Xuan Pan, Xiangrui Cai, Ying Zhang, and Xiaojie Yuan. 2024. Learning time slot preferences via mobility tree for next poi recommendation. Proceedings of the AAAI Conference on Artificial Intelligence, 38(8):8535–8543. find
  13. Yaxuan Kong, Yiyuan Yang, Yoontae Hwang, Wenjie Du, Stefan Zohren, Zhangyang Wang, Ming Jin, and Qingsong Wen. 2025. Time-mqa: Time series multi-task question answering with context enhancement. arXiv preprint arXiv:2503.01875.
  14. Shengjie Lai, Alessandro Sorichetta, Jessica Steele, Corrine W Ruktanonchai, Alexander D Cunningham, Grant Rogers, Patrycja Koper, Dorothea Woods, Maksym Bondarenko, Nick W Ruktanonchai, and 1 others. 2022. Global holiday datasets for understanding seasonal human mobility and population dynamics. Sc… find
  15. Zhengxing Lan, Lingshan Liu, Bo Fan, Yisheng Lv, Yilong Ren, and Zhiyong Cui. 2024. Traj-llm: A new exploration for empowering trajectory prediction with pre-trained large language models. IEEE Transactions on Intelligent Vehicles. find
  16. Xuchuan Li, Fei Huang, Jianrong Lv, Zhixiong Xiao, Guolong Li, and Yang Yue. 2024. Be more real: Travel diary generation using llm agents and individual profiles. arXiv preprint arXiv:2407.18932. — [arXiv:2407.18932](https://arxiv.org/abs/2407.18932) find
  17. Chenxi Liu, Zhu Xiao, Wangchen Long, Tong Li, Hongbo Jiang, and Keqin Li. 2025a. Vehicle trajectory data processing, analytics, and applications: A survey. ACM Computing Surveys, 57(9):1–36. find
  18. Tianming Liu, Jirong Yang, and Yafeng Yin. 2025b. Toward llm-agent-based modeling of transportation systems: A conceptual framework. Artificial Intelligence for Transportation, 1:100001. find
  19. Yang Liu, Jiahuan Cao, Chongyu Liu, Kai Ding, and Lianwen Jin. 2025c. Datasets for large language models: A comprehensive survey. Artificial Intelligence Review, 58(12):403. find
  20. Joseph Molloy, Alberto Castro, Thomas Götschi, Beaumont Schoeman, Christopher Tchervenkov, Uros Tomic, Beat Hintermann, and Kay W Axhausen. 2023. The mobis dataset: a large gps dataset of mobility behaviour in switzerland. Transportation, 50(5):1983–2007. find
  21. Hamed Nilforoshan, Wenli Looi, Emma Pierson, Blanca Villanueva, Nic Fishman, Yiling Chen, John Sholar, Beth Redbird, David Grusky, and Jure Leskovec. 2023. Human mobility networks reveal increased segregation in large cities. Nature, 624(7992):586– 592. find
  22. Guoying Qiu, Guoming Tang, Chuandong Li, Deke Guo, Yulong Shen, and Yan Gan. 2023. Behavioral-semantic privacy protection for continual social mobility in mobile-internet services. IEEE Internet of Things Journal, 11(1):462–477. find
  23. Benjamin Reichman, Xiaofan Yu, Lanxiang Hu, Jack Truxal, Atishay Jain, Rushil Chandrupatla, Tajana S Rosing, and Larry Heck. 2025. Sensorqa: A question answering benchmark for daily-life monitoring. In Proceedings of the 23rd ACM Conference on Embedded Networked Sensor Systems, pages 282–289. find
  24. Zijian Shao, Jiancan Wu, Weijian Chen, and Xiang Wang. 2025. Personal travel solver: A preference-driven LLM-solver system for travel planning. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 27622–27642, Vienna, Austria. Asso… find
  25. Xuan Shen, Hangyu Zheng, Yifan Gong, Zhenglun Kong, Changdi Yang, Zheng Zhan, Yushu Wu, Xue Lin, Yanzhi Wang, Pu Zhao, and 1 others. 2025. Sparse learning for state space models on mobile. In The Thirteenth International Conference on Learning Representations. find
  26. Chris Stanford, Suman Adari, Xishun Liao, Yueshuai He, Qinhua Jiang, Chenchen Kuai, Jiaqi Ma, Emmanuel Tung, Yinlong Qian, Lingyi Zhao, and 1 others. 2024. Numosim: A synthetic mobility dataset with anomaly detection benchmarks. In Proceedings of the 1st ACM SIGSPATIAL International Workshop on Geos… find
  27. Jiaan Wang, Fandong Meng, Zengkui Sun, Yunlong Liang, Yuxuan Cao, Jiarong Xu, Haoxiang Shi, and Jie Zhou. 2025. An empirical study of many-to-many summarization with large language models. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Pape… find
  28. Jiawei Wang, Renhe Jiang, Chuang Yang, Zengqing Wu, Makoto Onizuka, Ryosuke Shibasaki, Noboru Koshizuka, and Chuan Xiao. 2024. Large language models as urban residents: An llm agent framework for personal mobility generation. Advances in Neural Information Processing Systems, 37:124547–124574. find
  29. Xinglei Wang, Meng Fang, Zichao Zeng, and Tao Cheng. 2023. Where would i go next? large language models as human mobility predictors. arXiv preprint arXiv:2308.15197. — [arXiv:2308.15197](https://arxiv.org/abs/2308.15197) find
  30. Junde Wu, Jiayuan Zhu, Yuyuan Liu, Min Xu, and Yueming Jin. 2025. Agentic reasoning: A streamlined framework for enhancing LLM reasoning with agentic tools. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 28489– 28503, Vienna,…
  31. Zhiwen Xiao and Huagang Tong. 2025. Federated contrastive learning with feature-based distillation for human activity recognition. IEEE Transactions on Computational Social Systems. find
  32. Zhu Xiao, Linshan Wu, Hongbo Jiang, Zheng Qin, Chengxi Gao, You Li, Hongyang Chen, and Jiangchuan Liu. 2023. Exploring intercity mobility in urban agglomeration: Evidence from private car trajectory data. IEEE Transactions on Computational Social Systems, 11(2):2940–2954. find
  33. Takahiro Yabe, Kota Tsubouchi, Toru Shimizu, Yoshihide Sekimoto, Kaoru Sezaki, Esteban Moro, and Alex Pentland. 2024. Yjmob100k: City-scale and longitudinal dataset of anonymized human mobility trajectories. Scientific Data, 11(1):397. find
  34. Kairui Yang, Zihao Guo, Gengjie Lin, Haotian Dong, Zhao Huang, Yipeng Wu, Die Zuo, Jibin Peng, Ziyuan Zhong, Xin WANG, Qing Guo, Xiaosong Jia, Junchi Yan, and Di Lin. 2025a. Trajectory-LLM: A language-based data generator for trajectory prediction in autonomous driving. In The Thirteenth Internation… find
  35. Sean Bin Yang, Ying Sun, Yunyao Cheng, Yan Lin, Kristian Torp, and Jilin Hu. 2025b. Spatio-temporal trajectory foundation model-recent advances and future directions. arXiv preprint arXiv:2511.20729. — [arXiv:2511.20729](https://arxiv.org/abs/2511.20729) find
  36. Zhiwei Yin, Bin Jia, Xiao-Yong Yan, Yitao Yang, Hao Ji, and Ziyou Gao. 2026. Classification of the freight trip purpose of heavy trucks using trajectory data and waybill data. Transportation Research Part E: Logistics and Transportation Review, 206:104584. find
  37. Ayda Zaroujtaghi, Omid Mansourihanis, Mohammad Tayarani, Fatemeh Mansouri, Moein Hemmati, and Ali Soltani. 2025. A systematic review of gis evolution in transportation planning: Towards ai integration. Future Transportation, 5(3):97. find
  38. Meijing Zhang and Ying Xu. 2025. Transmode-llm: Feature-informed natural language modeling with domain-enhanced prompting for travel behavior modeling. In Large Language Models for Scientific and Societal Advances. find
  39. Yu Zheng, Hao Fu, Xing Xie, Wei-Ying Ma, and Quan-nan Li. 2011. Geolife GPS trajectory dataset - User Guide, geolife gps trajectories 1.1 edition. find
  40. Yuanshao Zhu, Yongchao Ye, Ying Wu, Xiangyu Zhao, and James Yu. 2023. Synmob: Creating high-fidelity synthetic gps trajectory dataset for urban mobility analysis. Advances in Neural Information Processing Systems, 36:22961–22977. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/zhen-2026-travelbehaviorqa-benchmark-dataset-behavioral-interpretation/