时空数据(Spatial-temporal)论文总结（时空预测，轨迹数据，人群移动，天气预报，多)
KDD 2026将在2026年8月9日至13日于韩国济州Jeju, Korea 举行。本文总结了KDD 2026July Cycle上有关时空数据Spatial-Temporal的相关论文。时空数据Topic时空交通预测轨迹数据挖掘表示生成人群移动天气预报以及多模态大模型和Agent在时空数据的应用等。Research Track1. Incident-Guided Spatiotemporal Traffic Forecasting.2. Leveraging the Spatial Hierarchy: Coarse-to-fine Trajectory Generation via Cascaded Hybrid Diffusion.3. DSPIGCN: Dual-stream Physics-informed Graph Convolutional Network for Reliable Pedestrian Trajectory Prediction.4. CFLight: Enhancing Safety with Traffic Signal Control through Counterfactual Learning.5. Multi-View Urban Region Embedding via Commonality-Specificity Disentanglement.6. Traj-MLLM: Can Multimodal Large Language Models Reform Trajectory Data Mining?7. UniLLM: A Unified Large Language Model for Multi-Modal Urban Dynamics Prediction.8. RIPCN: A Road Impedance Principal Component Network for Probabilistic Traffic Flow Forecasting.9. UniExtreme: A Universal Foundation Model for Extreme Weather Forecasting.10. AnchorGK: Anchor-based Incremental and Stratified Graph Learning Framework for Inductive Spatio-Temporal Kriging.11. Beyond Routines: Adaptive Mobility Prediction via Sequential-Relational Fusion.12. Towards Resilient Transportation: A Conditional Transformer for Accident-Informed Traffic Forecasting.13. CSSG: A Continuous Spatio-temporal Graph Learning Framework with Scalable Spatial Granularity.14. MoST: A Foundation Model for Multi-modality Spatio-temporal Traffic Prediction.15. FaST: Efficient and Effective Long-Horizon Forecasting for Large-Scale Spatial-Temporal Graphs via Mixture-of-Experts.16. Think2Go: Generative Next POI Recommendation with LLM Reasoning.17. KFTD: Koopman-Fourier Time-Differentiable Network for Continuous Ocean Spatiotemporal Forecasting.18. Spatiotemporal Graph Learning with Direct Volumetric Information Passing and Feature Enhancement.ADS Track19. StormMind: Disentangled Layerwise Modeling for Convective Weather Systems.20. ASTAFN: Bridging the Gap Between Weather Foundation Models and Accurate Station-Level Forecasting.Data Benchmark Track21. Generating Realistic Human Mobility Data with Hybrid Large Language Model Agent.22. Learning Multimodal Embeddings for Traffic Accident Prediction and Causal Estimation.23. FireSentry: A Multi-Modal Spatio-temporal Benchmark Dataset for Fine-Grained Wildfire Spread Forecasting.Research Track1 Incident-Guided Spatiotemporal Traffic Forecasting.链接https://dl.acm.org/doi/10.1145/3770854.3780215代码https://github.com/fanlixiang/IGSTGNN作者Lixiang Fan, Bohao Li, Tao Zou, Junchen Ye, Bowen Du关键词交通预测事件驱动2 Leveraging the Spatial Hierarchy: Coarse-to-fine Trajectory Generation via Cascaded Hybrid Diffusion.链接https://dl.acm.org/doi/10.1145/3770854.3780191代码https://github.com/urban-mobility-generation/Cardiff作者Baoshen Guo, Zhiqing Hong, Junyi Li, Shenhao Wang, Jinhua Zhao关键词轨迹生成扩散模型多层级多尺度3 DSPIGCN: Dual-stream Physics-informed Graph Convolutional Network for Reliable Pedestrian Trajectory Prediction.链接https://dl.acm.org/doi/10.1145/3770854.3780329作者Runkang Guo, Bin Chen, Zhengqiu Zhu, Chen Gao, Yong Zhao, Quanjun Yin关键词轨迹预测物理驱动对偶时空图物理运动学锚点采样?4 CFLight: Enhancing Safety with Traffic Signal Control through Counterfactual Learning.链接https://dl.acm.org/doi/10.1145/3770854.3780308代码https://github.com/MJLee00/CFLight作者Mingyuan Li, Chunyu Liu, Zhuojun Li, Xiao Liu, Guangsheng Yu, Bo Du, Jun Shen, Qiang Wu关键词信号灯控制反事实学习5 Multi-View Urban Region Embedding via Commonality-Specificity Disentanglement.链接https://dl.acm.org/doi/10.1145/3770854.3780234代码https://github.com/AIMUrban/ComSRE作者Zechen Li, Hongwei Jia, Kai Zhao, Weiming Huang, Meng Chen关键词城市区域嵌入多视图表示学习城市画像6 Traj-MLLM: Can Multimodal Large Language Models Reform Trajectory Data Mining?链接https://dl.acm.org/doi/10.1145/3770854.3780225作者Shuo Liu, Di Yao, Yan Lin, Gao Cong, Jingping Bi关键词轨迹表示学习多模态大模型7 UniLLM: A Unified Large Language Model for Multi-Modal Urban Dynamics Prediction.链接https://dl.acm.org/doi/10.1145/3770854.3780232代码https://github.com/Yliu1111/UniLLM作者Yuhang Liu, Yingxue Zhang, Xin Zhang, Yanhua Li, Jun Luo关键词城市需求预测多模态大模型统一模型8 RIPCN: A Road Impedance Principal Component Network for Probabilistic Traffic Flow Forecasting.链接https://dl.acm.org/doi/10.1145/3770854.3780183代码https://github.com/LvHaochenBANG/RIPCN作者Haochen Lv, Yan Lin, Shengnan Guo, Xiaowei Mao, Hong Nie, Letian Gong, Youfang Lin, Huaiyu Wan关键词交通预测不确定性估计主成分分析9 UniExtreme: A Universal Foundation Model for Extreme Weather Forecasting.链接https://dl.acm.org/doi/10.1145/3770854.3780172代码https://github.com/usail-hkust/UniExtreme作者Hang Ni, Weijia Zhang, Hao Liu关键词极端天气预测基础模型频域10 AnchorGK: Anchor-based Incremental and Stratified Graph Learning Framework for Inductive Spatio-Temporal Kriging.链接https://dl.acm.org/doi/10.1145/3770854.3780189代码https://github.com/xren451/Spatial-interpolation作者Xiaobin Ren, Kaiqi Zhao, Katerina Taskova, Patricia Riddle关键词时空克里格增量训练不完美特征空间相关性GNN11 Beyond Routines: Adaptive Mobility Prediction via Sequential-Relational Fusion.链接https://dl.acm.org/doi/10.1145/3770854.3780268代码https://github.com/AIMUrban/ROAM作者Tianao Sun, Ruizhe Liu, Wenzhen Jia, Kai Zhao, Weiming Huang, Meng Chen关键词下一位置预测非常规行为人类移动建模12 Towards Resilient Transportation: A Conditional Transformer for Accident-Informed Traffic Forecasting.链接https://dl.acm.org/doi/10.1145/3770854.3780312代码https://github.com/Dreamzz5/ConFormer作者Hongjun Wang, Jiawei Yong, Jiawei Wang, Shintaro Fukushima, Renhe Jiang关键词时空预测交通事故13 CSSG: A Continuous Spatio-temporal Graph Learning Framework with Scalable Spatial Granularity.链接https://dl.acm.org/doi/10.1145/3770854.3780161代码https://github.com/kaiwxai/CSSG作者Kaiwen Xia, Li Lin, Qi Zhang, Xinrui Zhang, Shuai Wang, Xuming Hu, Philip S. Yu关键词时空预测动态图可扩展的空间粒度14 MoST: A Foundation Model for Multi-modality Spatio-temporal Traffic Prediction.链接https://dl.acm.org/doi/10.1145/3770854.3780162作者Ronghui Xu, Jihao Chen, Jindong Tian, Chenjuan Guo, Bin Yang关键词时空预测多模态学习基础模型15 FaST: Efficient and Effective Long-Horizon Forecasting for Large-Scale Spatial-Temporal Graphs via Mixture-of-Experts.链接https://dl.acm.org/doi/10.1145/3770854.3780165代码https://github.com/yijizhao/FaST作者Yiji Zhao, Zihao Zhong, Ao Wang, Haomin Wen, Ming Jin, Yuxuan Liang, Huaiyu Wan, Hao Wu关键词长程预测大规模时空图混合专家系统MoE)16 Think2Go: Generative Next POI Recommendation with LLM Reasoning.链接https://dl.acm.org/doi/10.1145/3770854.3780334作者Zhuang Zhuang, Shanshan Feng, Hangwei Qian, Mingqi Yang, Heng Qi, Yanming Shen, Baocai Yin关键词POI推荐生成模型推理模型LLM17 KFTD: Koopman-Fourier Time-Differentiable Network for Continuous Ocean Spatiotemporal Forecasting.链接https://dl.acm.org/doi/10.1145/3770854.3780278作者Qinghui Chen, Zekai Zhang, Hailong Liu, Jinglin Zhang, Cong Bai关键词时空预报海洋动力学业务海洋预报物理信息学习库普曼神经网络算子18 Spatiotemporal Graph Learning with Direct Volumetric Information Passing and Feature Enhancement.链接https://dl.acm.org/doi/10.1145/3770854.3780155代码https://github.com/intell-sci-comput/CeFeGNN作者Yuan Mi, Qi Wang, Xueqin Hu, Yike Guo, Ji-Rong Wen, Yang Liu, Hao Sun关键词图学习时空预测高阶动力学ADS Track19 StormMind: Disentangled Layerwise Modeling for Convective Weather Systems.链接https://dl.acm.org/doi/10.1145/3770854.3783926作者Jun Chen, Minghui Qiu, Lin Chen, Yan Fang, Shuxin Zhong, Binghong Chen, Kaishun Wu关键词天气系统、降水临近预报20 ASTAFN: Bridging the Gap Between Weather Foundation Models and Accurate Station-Level Forecasting.链接https://dl.acm.org/doi/10.1145/3770854.3783935代码https://github.com/xubihe-bjtu/ASTAFN作者Bihe Xu, Zhicheng Yan, Qingyong Li, Zhiqing Guo, Dong Zheng, Wen Yao, Bo Wang, Zhao Wang, Yangliao Geng关键词天气基础模型站点级天气预报Data Benchmark Track21 Generating Realistic Human Mobility Data with Hybrid Large Language Model Agent.链接https://dl.acm.org/doi/10.1145/3770854.3785685代码https://github.com/tsinghua-fib-lab/CoPB作者Chenyang Shao, Bingbing Fan, Jingtao Ding, Yuan Yuan, Meng Wang, Fengli Xu关键词移动数据生成扩散模型LLM Agent22 Learning Multimodal Embeddings for Traffic Accident Prediction and Causal Estimation.链接https://dl.acm.org/doi/10.1145/3770854.3785677代码https://github.com/VirtuosoResearch/MMTraCE作者Ziniu Zhang, Minxuan Duan, Haris N. Koutsopoulos, Hongyang R. Zhang关键词多模态学习卫星图像道路安全因果分析21 FireSentry: A Multi-Modal Spatio-temporal Benchmark Dataset for Fine-Grained Wildfire Spread Forecasting.链接https://dl.acm.org/doi/10.1145/3770854.3785696代码https://github.com/Munan222/FireSentry-Benchmark-Dataset作者Nan Zhou, Huandong Wang, Jiahao Li, Han Li, Yali Song, Qiuhua Wang, Yong Li, Xinlei Chen关键词野火蔓延预测精细化多模态数据集生成模型 KDD 2026 | (7月轮)时空数据(Spatial-temporal)论文总结（时空预测，轨迹数据，人群移动，天气预报，多
KDD 2026将在2026年8月9日至13日于韩国济州Jeju, Korea 举行。本文总结了KDD 2026July Cycle上有关时空数据Spatial-Temporal的相关论文。时空数据Topic时空交通预测轨迹数据挖掘表示生成人群移动天气预报以及多模态大模型和Agent在时空数据的应用等。Research Track1. Incident-Guided Spatiotemporal Traffic Forecasting.2. Leveraging the Spatial Hierarchy: Coarse-to-fine Trajectory Generation via Cascaded Hybrid Diffusion.3. DSPIGCN: Dual-stream Physics-informed Graph Convolutional Network for Reliable Pedestrian Trajectory Prediction.4. CFLight: Enhancing Safety with Traffic Signal Control through Counterfactual Learning.5. Multi-View Urban Region Embedding via Commonality-Specificity Disentanglement.6. Traj-MLLM: Can Multimodal Large Language Models Reform Trajectory Data Mining?7. UniLLM: A Unified Large Language Model for Multi-Modal Urban Dynamics Prediction.8. RIPCN: A Road Impedance Principal Component Network for Probabilistic Traffic Flow Forecasting.9. UniExtreme: A Universal Foundation Model for Extreme Weather Forecasting.10. AnchorGK: Anchor-based Incremental and Stratified Graph Learning Framework for Inductive Spatio-Temporal Kriging.11. Beyond Routines: Adaptive Mobility Prediction via Sequential-Relational Fusion.12. Towards Resilient Transportation: A Conditional Transformer for Accident-Informed Traffic Forecasting.13. CSSG: A Continuous Spatio-temporal Graph Learning Framework with Scalable Spatial Granularity.14. MoST: A Foundation Model for Multi-modality Spatio-temporal Traffic Prediction.15. FaST: Efficient and Effective Long-Horizon Forecasting for Large-Scale Spatial-Temporal Graphs via Mixture-of-Experts.16. Think2Go: Generative Next POI Recommendation with LLM Reasoning.17. KFTD: Koopman-Fourier Time-Differentiable Network for Continuous Ocean Spatiotemporal Forecasting.18. Spatiotemporal Graph Learning with Direct Volumetric Information Passing and Feature Enhancement.ADS Track19. StormMind: Disentangled Layerwise Modeling for Convective Weather Systems.20. ASTAFN: Bridging the Gap Between Weather Foundation Models and Accurate Station-Level Forecasting.Data Benchmark Track21. Generating Realistic Human Mobility Data with Hybrid Large Language Model Agent.22. Learning Multimodal Embeddings for Traffic Accident Prediction and Causal Estimation.23. FireSentry: A Multi-Modal Spatio-temporal Benchmark Dataset for Fine-Grained Wildfire Spread Forecasting.Research Track1 Incident-Guided Spatiotemporal Traffic Forecasting.链接https://dl.acm.org/doi/10.1145/3770854.3780215代码https://github.com/fanlixiang/IGSTGNN作者Lixiang Fan, Bohao Li, Tao Zou, Junchen Ye, Bowen Du关键词交通预测事件驱动2 Leveraging the Spatial Hierarchy: Coarse-to-fine Trajectory Generation via Cascaded Hybrid Diffusion.链接https://dl.acm.org/doi/10.1145/3770854.3780191代码https://github.com/urban-mobility-generation/Cardiff作者Baoshen Guo, Zhiqing Hong, Junyi Li, Shenhao Wang, Jinhua Zhao关键词轨迹生成扩散模型多层级多尺度3 DSPIGCN: Dual-stream Physics-informed Graph Convolutional Network for Reliable Pedestrian Trajectory Prediction.链接https://dl.acm.org/doi/10.1145/3770854.3780329作者Runkang Guo, Bin Chen, Zhengqiu Zhu, Chen Gao, Yong Zhao, Quanjun Yin关键词轨迹预测物理驱动对偶时空图物理运动学锚点采样?4 CFLight: Enhancing Safety with Traffic Signal Control through Counterfactual Learning.链接https://dl.acm.org/doi/10.1145/3770854.3780308代码https://github.com/MJLee00/CFLight作者Mingyuan Li, Chunyu Liu, Zhuojun Li, Xiao Liu, Guangsheng Yu, Bo Du, Jun Shen, Qiang Wu关键词信号灯控制反事实学习5 Multi-View Urban Region Embedding via Commonality-Specificity Disentanglement.链接https://dl.acm.org/doi/10.1145/3770854.3780234代码https://github.com/AIMUrban/ComSRE作者Zechen Li, Hongwei Jia, Kai Zhao, Weiming Huang, Meng Chen关键词城市区域嵌入多视图表示学习城市画像6 Traj-MLLM: Can Multimodal Large Language Models Reform Trajectory Data Mining?链接https://dl.acm.org/doi/10.1145/3770854.3780225作者Shuo Liu, Di Yao, Yan Lin, Gao Cong, Jingping Bi关键词轨迹表示学习多模态大模型7 UniLLM: A Unified Large Language Model for Multi-Modal Urban Dynamics Prediction.链接https://dl.acm.org/doi/10.1145/3770854.3780232代码https://github.com/Yliu1111/UniLLM作者Yuhang Liu, Yingxue Zhang, Xin Zhang, Yanhua Li, Jun Luo关键词城市需求预测多模态大模型统一模型8 RIPCN: A Road Impedance Principal Component Network for Probabilistic Traffic Flow Forecasting.链接https://dl.acm.org/doi/10.1145/3770854.3780183代码https://github.com/LvHaochenBANG/RIPCN作者Haochen Lv, Yan Lin, Shengnan Guo, Xiaowei Mao, Hong Nie, Letian Gong, Youfang Lin, Huaiyu Wan关键词交通预测不确定性估计主成分分析9 UniExtreme: A Universal Foundation Model for Extreme Weather Forecasting.链接https://dl.acm.org/doi/10.1145/3770854.3780172代码https://github.com/usail-hkust/UniExtreme作者Hang Ni, Weijia Zhang, Hao Liu关键词极端天气预测基础模型频域10 AnchorGK: Anchor-based Incremental and Stratified Graph Learning Framework for Inductive Spatio-Temporal Kriging.链接https://dl.acm.org/doi/10.1145/3770854.3780189代码https://github.com/xren451/Spatial-interpolation作者Xiaobin Ren, Kaiqi Zhao, Katerina Taskova, Patricia Riddle关键词时空克里格增量训练不完美特征空间相关性GNN11 Beyond Routines: Adaptive Mobility Prediction via Sequential-Relational Fusion.链接https://dl.acm.org/doi/10.1145/3770854.3780268代码https://github.com/AIMUrban/ROAM作者Tianao Sun, Ruizhe Liu, Wenzhen Jia, Kai Zhao, Weiming Huang, Meng Chen关键词下一位置预测非常规行为人类移动建模12 Towards Resilient Transportation: A Conditional Transformer for Accident-Informed Traffic Forecasting.链接https://dl.acm.org/doi/10.1145/3770854.3780312代码https://github.com/Dreamzz5/ConFormer作者Hongjun Wang, Jiawei Yong, Jiawei Wang, Shintaro Fukushima, Renhe Jiang关键词时空预测交通事故13 CSSG: A Continuous Spatio-temporal Graph Learning Framework with Scalable Spatial Granularity.链接https://dl.acm.org/doi/10.1145/3770854.3780161代码https://github.com/kaiwxai/CSSG作者Kaiwen Xia, Li Lin, Qi Zhang, Xinrui Zhang, Shuai Wang, Xuming Hu, Philip S. Yu关键词时空预测动态图可扩展的空间粒度14 MoST: A Foundation Model for Multi-modality Spatio-temporal Traffic Prediction.链接https://dl.acm.org/doi/10.1145/3770854.3780162作者Ronghui Xu, Jihao Chen, Jindong Tian, Chenjuan Guo, Bin Yang关键词时空预测多模态学习基础模型15 FaST: Efficient and Effective Long-Horizon Forecasting for Large-Scale Spatial-Temporal Graphs via Mixture-of-Experts.链接https://dl.acm.org/doi/10.1145/3770854.3780165代码https://github.com/yijizhao/FaST作者Yiji Zhao, Zihao Zhong, Ao Wang, Haomin Wen, Ming Jin, Yuxuan Liang, Huaiyu Wan, Hao Wu关键词长程预测大规模时空图混合专家系统MoE)16 Think2Go: Generative Next POI Recommendation with LLM Reasoning.链接https://dl.acm.org/doi/10.1145/3770854.3780334作者Zhuang Zhuang, Shanshan Feng, Hangwei Qian, Mingqi Yang, Heng Qi, Yanming Shen, Baocai Yin关键词POI推荐生成模型推理模型LLM17 KFTD: Koopman-Fourier Time-Differentiable Network for Continuous Ocean Spatiotemporal Forecasting.链接https://dl.acm.org/doi/10.1145/3770854.3780278作者Qinghui Chen, Zekai Zhang, Hailong Liu, Jinglin Zhang, Cong Bai关键词时空预报海洋动力学业务海洋预报物理信息学习库普曼神经网络算子18 Spatiotemporal Graph Learning with Direct Volumetric Information Passing and Feature Enhancement.链接https://dl.acm.org/doi/10.1145/3770854.3780155代码https://github.com/intell-sci-comput/CeFeGNN作者Yuan Mi, Qi Wang, Xueqin Hu, Yike Guo, Ji-Rong Wen, Yang Liu, Hao Sun关键词图学习时空预测高阶动力学ADS Track19 StormMind: Disentangled Layerwise Modeling for Convective Weather Systems.链接https://dl.acm.org/doi/10.1145/3770854.3783926作者Jun Chen, Minghui Qiu, Lin Chen, Yan Fang, Shuxin Zhong, Binghong Chen, Kaishun Wu关键词天气系统、降水临近预报20 ASTAFN: Bridging the Gap Between Weather Foundation Models and Accurate Station-Level Forecasting.链接https://dl.acm.org/doi/10.1145/3770854.3783935代码https://github.com/xubihe-bjtu/ASTAFN作者Bihe Xu, Zhicheng Yan, Qingyong Li, Zhiqing Guo, Dong Zheng, Wen Yao, Bo Wang, Zhao Wang, Yangliao Geng关键词天气基础模型站点级天气预报Data Benchmark Track21 Generating Realistic Human Mobility Data with Hybrid Large Language Model Agent.链接https://dl.acm.org/doi/10.1145/3770854.3785685代码https://github.com/tsinghua-fib-lab/CoPB作者Chenyang Shao, Bingbing Fan, Jingtao Ding, Yuan Yuan, Meng Wang, Fengli Xu关键词移动数据生成扩散模型LLM Agent22 Learning Multimodal Embeddings for Traffic Accident Prediction and Causal Estimation.链接https://dl.acm.org/doi/10.1145/3770854.3785677代码https://github.com/VirtuosoResearch/MMTraCE作者Ziniu Zhang, Minxuan Duan, Haris N. Koutsopoulos, Hongyang R. Zhang关键词多模态学习卫星图像道路安全因果分析21 FireSentry: A Multi-Modal Spatio-temporal Benchmark Dataset for Fine-Grained Wildfire Spread Forecasting.链接https://dl.acm.org/doi/10.1145/3770854.3785696代码https://github.com/Munan222/FireSentry-Benchmark-Dataset作者Nan Zhou, Huandong Wang, Jiahao Li, Han Li, Yali Song, Qiuhua Wang, Yong Li, Xinlei Chen关键词野火蔓延预测精细化多模态数据集生成模型