基于大语言模型的时空数据零样本插补

梅雅欣; 秦慧玲; 梁玉珠; 张广学; 王田

doi:10.12263/DZXB.20250473

您当前的位置：

首页 >

文章列表页 >

基于大语言模型的时空数据零样本插补

大模型与互联网 | 更新时间：2025-12-27

- 基于大语言模型的时空数据零样本插补
- LLM-Based Zero-Shot Imputation of Spatiotemporal Data
- 电子学报 2025年53卷第9期页码：3047-3059
- 作者机构：
  
  1.北京师范大学人工智能与未来网络研究院，广东珠海 519087
  2.北京师范大学人工智能学院，北京 100875
- 作者简介：
  
  [ "秦慧玲女，1996年8月出生于广西壮族自治区桂林市.现为北京师范大学珠海校区讲师、博士生导师.主要研究方向为时空数据挖掘、城市计算.主持国家级科研项目1项，在国内外发表顶级学术论文十余篇.中国电子学会会员编号：E190188248M.E-mail: qinhuiling@bnu.edu.cn" ]
  [ "梁玉珠男，1994年10月出生于安徽省亳州市.2025年毕业于北京师范大学人工智能学院，现为北京师范大学博士后.E-mail: yuzhuliang@bnu.edu.cn" ]
  [ "张广学男，1992年1月出生于河南省开封市.现为北京师范大学珠海校区自然科学高等研究院工程师.主要研究方向为声纹识别、语音识别、边缘智能.E-mail: guangxue_zhang@bnu.edu.cn" ]
  [ "王田男，1982年2月出生于湖南省汨罗市.现为北京师范大学长聘教授、博士生导师，国家级青年拔尖人才.主要研究方向为物联网、边缘智能.中国电子学会会员编号：E190026040M.E-mail: tianwang@bnu.edu.cn" ]
- 基金信息：
  
  国家自然科学基金(62372047);北京市自然科学基金(4232028);珠海市产学研项目(2220004002686;2320004002812);珠海市基础与应用共础课题研究项目(2220004002619);珠海市社会发展领域科技计划项目(2320004000213)
- DOI：10.12263/DZXB.20250473
  中图分类号： TP399;
- 收稿：2025-06-05，
  
  录用：2025-09-19，
  
  纸质出版：2025-09-25
- 稿件说明：
移动端阅览
梅雅欣, 秦慧玲, 梁玉珠, 等. 基于大语言模型的时空数据零样本插补[J]. 电子学报, 2025, 53(09): 3047-3059.

MEI Ya-xin, QIN Hui-ling, LIANG Yu-zhu, et al. LLM-Based Zero-Shot Imputation of Spatiotemporal Data[J]. Acta Electronica Sinica, 2025, 53(09): 3047-3059.
梅雅欣, 秦慧玲, 梁玉珠, 等. 基于大语言模型的时空数据零样本插补[J]. 电子学报, 2025, 53(09): 3047-3059. DOI：10.12263/DZXB.20250473

MEI Ya-xin, QIN Hui-ling, LIANG Yu-zhu, et al. LLM-Based Zero-Shot Imputation of Spatiotemporal Data[J]. Acta Electronica Sinica, 2025, 53(09): 3047-3059. DOI：10.12263/DZXB.20250473

摘要

物联网感知数据由于部署成本、环境约束、设备故障等多重因素普遍存在数据稀疏问题，严重制约智能感知系统的整体性能.现有插补方法大多依赖标注数据进行监督训练，在面对新场景“冷启动”时泛化能力严重不足，难以满足物联网快速部署和跨域应用的实际需求.本文首次将大语言模型的内在推理能力引入时空数据插补领域，提出了基于多智能体协同推理的ZeroImpute框架，实现了从传统“数据驱动学习”向“知识驱动推理”的范式转换.该方法构建了由专门化任务的大语言模型智能体组成的协同推理系统：时序分析智能体负责复杂时间依赖关系的语义理解与推理，通过双向序列建模捕获前向演化趋势和后向约束条件；空间分析智能体专注于动态空间关系的建模与解析，基于时序上下文指导实现时变空间相关性的精准识别；插补决策智能体整合多源语义知识，运用自适应权重融合算法完成最终的智能插补决策.各智能体通过语义化的知识表达与逻辑推理实现对复杂时空模式的深度理解，将传统的数值计算问题转化为多智能体可协同处理的语义推理任务，突破了单一模型处理复杂时空关系的局限性.该框架具备一定的技术优势：首先，实现了真正的零样本泛化能力，无需任何领域特定的训练数据即可直接部署；其次，通过多智能体分工，提升了复杂时空模式的识别精度和推理质量；再次，具备良好的可解释性，智能体推理过程透明化，增强了系统的可信度；最后，即插即用部署大幅降低了实际应用的技术门槛和部署成本.在三个真实物联网数据集的综合评测中，ZeroImpute在完全零样本、零训练的严格设置下相比最优的专用深度学习模型在平均绝对误差MAE （Mean Absolute Error）指标上实现了至少4.5%的性能提升.此外，该方法在不同缺失率场景下均展现出鲁棒性，能够有效解决新部署区域快速上线、跨域数据插补泛化以及资源受限环境高效部署等关键实际问题.本研究开辟了多智能体协同推理的时空计算新范式，为时空数据插补领域提供了全新的技术路径，为推动物联网技术在更广泛领域的产业化应用提供了关键的技术支撑和理论基础.

Abstract

Internet of things (IoT) sensing data commonly suffers from data sparsity issues due to multiple factors including deployment costs

environmental constraints

and equipment failures

severely limiting the overall performance of intelligent sensing systems. Most existing imputation methods rely on labeled data for supervised training

resulting in severely insufficient generalization capabilities when facing “cold start” scenarios in new environments

failing to meet the practical demands of rapid IoT deployment and cross-domain applications. This paper introduces

for the first time

the intrinsic reasoning capabilities of large language models (LLM) into the spatiotemporal data imputation domain

proposing the ZeroImpute framework based on multi-agent collaborative reasoning that achieves a paradigmatic shift from traditional “data-driven learning” to “knowledge-driven reasoning.” The core innovation of this method lies in constructing a collaborative reasoning system comprising specialized task-oriented LLM agents: the temporal analysis agent is responsible for semantic understanding and reasoning of complex temporal dependencies

capturing forward evolutionary trends and backward constraint conditions through bidirectional sequence modeling; the spatial analysis agent focuses on modeling and parsing dynamic spatial relationships

achieving precise identification of time-varying spatial correlations through temporal context guidance; the imputation decision agent integrates multi-source semantic knowledge and employs adaptive weight fusion algorithms to complete final intelligent imputation decisions. Each agent achieves deep understanding of complex spatiotemporal patterns through semantic knowledge representation and logical reasoning

transforming traditional numerical computation problems into semantic reasoning tasks that can be collaboratively processed by multiple agents

thereby overcoming the limitations of single models in handling complex spatiotemporal relationships. The framework possesses significant technical advantages: first

it achieves true zero-shot generalization capability

enabling direct deployment without requiring any domain-specific training data; second

through multi-agent specialization

it enhances the identification accuracy and reasoning quality of complex spatiotemporal patterns; third

it exhibits excellent interpretability with transparent agent reasoning processes

enhancing system trustworthiness; finally

plug-and-play deployment substantially reduces technical barriers and deployment costs for practical applications. Comprehensive evaluations on three real-world IoT datasets demonstrate that ZeroImpute achieves at least a 4.5% performance improvement in MAE compared to the best-performing specialized deep learning models under strictly zero-shot

zero-training settings. Moreover

the method exhibits robustness across different missing rate scenarios

effectively addressing critical practical challenges including rapid deployment in new regions

cross-domain data imputation generalization

and efficient deployment in resource-constrained environments. This research pioneers a new paradigm of multi-agent collaborative reasoning for spatiotemporal computation

providing novel technical pathways for the spatiotemporal data imputation field and offering crucial technical support and theoretical foundations for advancing IoT technology adoption across broader industrial applications.

关键词

Keywords

references

张依琳 , 梁玉珠 , 尹沐君 , 等 . 移动边缘计算中计算卸载方案研究综述 [J ] . 计算机学报 , 2021 , 44 ( 12 ): 2406 - 2430 .

ZHANG Y L , LIANG Y Z , YIN M J , et al . Survey on the methods of computation of f loading in mobile edge computing [J ] . Chinese Journal of Computers , 2021 , 44 ( 12 ): 2406 - 2430 . (in Chinese)

申岩松 , 李琳 , 黄传明 . 全局和局部感知的交通速度预测模型 [J ] . 电子学报 , 2024 , 52 ( 9 ): 3195 - 3205 .

SHEN Y S , LI L , HUANG C M . Global and local information aware traffic speed prediction [J ] . Acta Electronica Sinica , 2024 , 52 ( 9 ): 3195 - 3205 . (in Chinese)

ZHANG J , TAO D C . Empowering things with intelligence: A survey of the progress, challenges, and opportunities in artificial intelligence of things [J ] . IEEE Internet of Things Journal , 2021 , 8 ( 10 ): 7789 - 7817 .

MEI Y X , WANG W H , LIANG Y Z , et al . Privacy-enhanced cooperative storage scheme for contact-free sensory data in AIoT with efficient synchronization [J ] . ACM Transactions on Sensor Networks , 2024 , 20 ( 4 ): 1 - 19 .

LUONG N C , HOANG D T , WANG P , et al . Data collection and wireless communication in Internet of Things (IoT) using economic analysis and pricing models: A survey [J ] . IEEE Communications Surveys & Tutorials , 2016 , 18 ( 4 ): 2546 - 2590 .

XU W Z , XIAO T , ZHANG J Q , et al . Minimizing the deployment cost of UAVs for delay-sensitive data collection in IoT networks [J ] . IEEE/ACM Transactions on Networking , 2022 , 30 ( 2 ): 812 - 825 .

GHDIRI O , JAAFAR W , ALFATTANI S , et al . Offline and online UAV-enabled data collection in time-constrained IoT networks [J ] . IEEE Transactions on Green Communications and Networking , 2021 , 5 ( 4 ): 1918 - 1933 .

CAPPONI A , FIANDRINO C , KLIAZOVICH D , et al . A cost-effective distributed framework for data collection in cloud-based mobile crowd sensing architectures [J ] . IEEE Transactions on Sustainable Computing , 2017 , 2 ( 1 ): 3 - 16 .

XU C F , GUO J X , LI Y P , et al . Dynamic parallel multi-server selection and allocation in collaborative edge computing [J ] . IEEE Transactions on Mobile Computing , 2024 , 23 ( 11 ): 10523 - 10537 .

蒋伟进 , 王海娟 , 周为 , 等 . 基于自适应连续时间的群智感知轨迹隐私保护方案 [J ] . 电子学报 , 2023 , 51 ( 10 ): 2894 - 2901 .

JIANG W J , WANG H J , ZHOU W , et al . Track privacy protection scheme based on adaptive continuous time in crowdsensing [J ] . Acta Electronica Sinica , 2023 , 51 ( 10 ): 2894 - 2901 . (in Chinese)

LI C X , LI Z T , LONG S Q , et al . Robust data inference and cost-effective cell selection for sparse mobile crowdsensing [J ] . IEEE/ACM Transactions on Networking , 2024 , 32 ( 5 ): 3760 - 3775 .

兰玉乾 , 饶元 , 李冠呈 , 等 . 基于内在质量约束的文本生成和评价综述 [J ] . 电子学报 , 2024 , 52 ( 2 ): 633 - 659 .

LAN Y Q , RAO Y , LI G C , et al . A survey of text generation and evaluation based on intrinsic quality constraints [J ] . Acta Electronica Sinica , 2024 , 52 ( 2 ): 633 - 659 . (in Chinese)

FENG M K , GU J J , QIU J , et al . From news to forecast: Integrating event analysis in LLM-based time series forecasting with reflection [C ] // Advances in Neural Information Processing Systems 37 . Vancouver : Neural Information Processing Systems Foundation, Inc. (NeurIPS) , 2024 : 58118 - 58153 .

YU H F , RAO N , DHILLON I S . Temporal regularized matrix factorization for high-dimensional time series prediction [C ] // Proceedings of the 30th International Conference on Neural Information Processing Systems . New York : ACM , 2016 : 847 - 855 .

CHEN X Y , HE Z C , CHEN Y X , et al . Missing traffic data imputation and pattern discovery with a Bayesian augmented tensor factorization model [J ] . Transportation Research Part C: Emerging Technologies , 2019 , 104 : 66 - 77 .

SALAKHUTDINOV R , MNIH A . Bayesian probabilistic matrix factorization using Markov chain Monte Carlo [C ] // Proceedings of the 25th International Conference on Machine Learning . New York : ACM , 2008 : 880 - 887 .

CHEN X Y , SUN L J . Bayesian temporal factorization for multidimensional time series prediction [J ] . IEEE Transactions on Pattern Analysis and Machine Intelligence , 2022 , 44 ( 9 ): 4659 - 4673 .

金红 , 胡智群 . 基于非负矩阵分解的稀疏网络社区发现算法 [J ] . 电子学报 , 2023 , 51 ( 10 ): 2950 - 2959 .

JIN H , HU Z Q . The non-negative matrix factorization based algorithm for community detection in sparse networks [J ] . Acta Electronica Sinica , 2023 , 51 ( 10 ): 2950 - 2959 . (in Chinese)

AYDILEK I B , ARSLAN A . A hybrid method for imputation of missing values using optimized fuzzy c -means with support vector regression and a genetic algorithm [J ] . Information Sciences , 2013 , 233 : 25 - 35 .

SHUMWAY R H , STOFFER D S . An approach to time series smoothing and forecasting using the em algorithm [J ] . Journal of Time Series Analysis , 1982 , 3 ( 4 ): 253 - 264 .

NELWAMONDO F V , MOHAMED S , MARWALA T . Missing data: A comparison of neural network and expectation maximization techniques [J ] . Current Science , 2007 , 93 ( 11 ): 1514 - 1521 .

CAO W , WANG D , LI J , et al . Brits: Bidirectional recurrent imputation for time series [C ] // Proceedings of the 32nd International Conference on Neural Information Processing Systems . New York : ACM , 2018 : 6776 - 6786 .

TASHIRO Y , SONG J M , SONG Y , et al . CSDI: Conditional score-based diffusion models for probabilistic time series imputation [C ] // Proceedings of the 35th International Conference on Neural Information Processing Systems . New York : ACM , 2021 : 24804 - 24816 .

MIAO X Y , WU Y Y , WANG J , et al . Generative semi-supervised learning for multivariate time series imputation [J ] . Proceedings of the AAAI Conference on Artificial Intelligence , 2021 , 35 ( 10 ): 8983 - 8991 .

LIU Y , WU H X , WANG J M , et al . Non-stationary transformers: Exploring the stationarity in time series forecasting [C ] // Proceedings of the 36th International Conference on Neural Information Processing Systems . New York : ACM , 2022 , 9881 - 9893 .

ZHANG Y H , YAN J C . Crossformer: Transformer utilizing cross-dimension dependency for multivariate time series forecasting [C ] // the 11th International Conference on Learning Representations . Appleton : ICLR , 2023 : 1 - 21 .

LIU Y , HU T G , ZHANG H R , et al . iTransformer: Inverted transformers are effective for time series forecasting [EB/OL ] . ( 2024-03-14 )[ 2025-05-25 ] . https://arXiv.org/abs/2310.06625 https://arXiv.org/abs/2310.06625 .

NIE T , QIN G Y , MA W , et al . ImputeFormer: Low rankness-induced transformers for generalizable spatiotemporal imputation [C ] // Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining . New York : ACM , 2024 : 2260 - 2271 .

DU W J , CÔTÉ D , LIU Y . SAITS: Self-attention-based imputation for time series [J ] . Expert Systems with Applications , 2023 , 219 : 119619 .

WU H X , HU T G , LIU Y , et al . TimesNet: Temporal 2D-variation modeling for general time series analysis [EB/OL ] . ( 2023-04-12 )[ 2025-05-25 ] . https://arXiv.org/abs/2210.02186 https://arXiv.org/abs/2210.02186 .

LUO D H , WANG X . Moderntcn: A modern pure convolution structure for general time series analysis [C ] // The 12th International Conference on Learning Representations . Appleton : ICLR , 2024 : 1 - 43 .

SONG C , LIN Y F , GUO S N , et al . Spatial-temporal synchronous graph convolutional networks: A new framework for spatial-temporal network data forecasting [J ] . Proceedings of the AAAI Conference on Artificial Intelligence , 2020 , 34 ( 1 ): 914 - 921 .

ZHANG Q R , HUANG C , XIA L H , et al . Automated spatio-temporal graph contrastive learning [C ] // Proceedings of the ACM Web Conference 2023 . New York : ACM , 2023 : 295 - 305 .

ZHAO L , SONG Y J , ZHANG C , et al . T-GCN: A temporal graph convolutional network for traffic prediction [J ] . IEEE Transactions on Intelligent Transportation Systems , 2020 , 21 ( 9 ): 3848 - 3858 .

CAO D F , WANG Y J , DUAN J Y , et al . Spectral temporal graph neural network for multivariate time-series forecasting [C ] // Proceedings of the 34th International Conference on Neural Information Processing Systems . New York : ACM , 2020 : 17766 - 17778 .

LIANG Z J , XU Y J , HONG Y F , et al . A survey of multimodel large language models [C ] // Proceedings of the 3rd International Conference on Computer, Artificial Intelligence and Control Engineering . New York : ACM , 2024 : 405 - 409 .

ZHOU T , NIU P , SUN L , et al . One fits all: Power general time series analysis by pretrained lm [C ] // Proceedings of the 37th International Conference on Neural Information Processing Systems . New York : ACM , 2023 : 43322 - 43355 .

JIN M , WANG S Y , MA L T , et al . Time-LLM: Time series forecasting by reprogramming large language models [EB/OL ] . ( 2024-01-29 )[ 2025-05-25 ] . https://arXiv.org/abs/2310.01728 https://arXiv.org/abs/2310.01728 .

XUE H , SALIM F D . PromptCast: A new prompt-based learning paradigm for time series forecasting [J ] . IEEE Transactions on Knowledge and Data Engineering , 2024 , 36 ( 11 ): 6851 - 6864 .

LI Z H , XIA L H , TANG J B , et al . UrbanGPT: Spatio-temporal large language models [C ] // Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining . New York : ACM , 2024 : 5351 - 5362 .

GRUVER N , FINZI M , QIU S , et al . Large language models are zero-shot time series forecasters [C ] // Proceedings of the 37th International Conference on Neural Information Processing Systems . New York : ACM , 2023 : 19622 - 19635 .

WANG J Y , JIANG J W , JIANG W J , et al . LibCity: An open library for traffic prediction [C ] // Proceedings of the 29th International Conference on Advances in Geographic Information Systems . New York : ACM , 2021 : 145 - 148 .

ZHENG Y , LIU F R , HSIEH H P . U-Air: When urban air quality inference meets big data [C ] // Proceedings of the 19th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining . New York : ACM , 2013 : 1436 - 1444 .

：作者简介：

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

面向LLM开放域问答中多方私有表格筛选：一种MPC可公开聚合审计与动态信誉的增强方法

基于共享知识模型的跨领域推荐算法

基于混杂社会网络的个性化Web服务推荐方法