Urban Trajectory Generation via Knowledge Graph-Enhanced Multi-Source Context Fusion

LI Kang; YU Juan; HAN Jian-min; QIU Sheng; YANG Qiong

doi:10.12263/DZXB.20250307

您当前的位置：

首页 >

文章列表页 >

Urban Trajectory Generation via Knowledge Graph-Enhanced Multi-Source Context Fusion

PAPERS | 更新时间：2026-02-05

- Urban Trajectory Generation via Knowledge Graph-Enhanced Multi-Source Context Fusion
- ACTA ELECTRONICA SINICA Vol. 53, Issue 10, Pages: 3551-3565(2025)
- 作者机构：
  
  1.浙江师范大学计算机与科学技术学院，浙江金华 321004
  2.浙江工业职业技术学院信息与设计学院，浙江绍兴 312099
- 作者简介：
- 基金信息：
  
  Humanities and Social Sciences Project of the Ministry of Education of China(22YJCZH215);Major Humanities and Social Sciences Research Projects in Zhejiang Higher Education Institutions(2023QN150);National Natural Science Foundation of China(61702148;61672648)
- DOI：10.12263/DZXB.20250307
  CLC： TP309.2;
- Received：21 April 2025，
  
  Accepted：10 October 2025，
  
  Published：25 October 2025
- 稿件说明：
移动端阅览
李康, 于娟, 韩建民, 等. 融合多源城市环境信息的知识图谱驱动轨迹生成模型[J]. 电子学报, 2025, 53(10): 3551-3565.

LI Kang, YU Juan, HAN Jian-min, et al. Urban Trajectory Generation via Knowledge Graph-Enhanced Multi-Source Context Fusion[J]. Acta Electronica Sinica, 2025, 53(10): 3551-3565.
李康, 于娟, 韩建民, 等. 融合多源城市环境信息的知识图谱驱动轨迹生成模型[J]. 电子学报, 2025, 53(10): 3551-3565. DOI：10.12263/DZXB.20250307

LI Kang, YU Juan, HAN Jian-min, et al. Urban Trajectory Generation via Knowledge Graph-Enhanced Multi-Source Context Fusion[J]. Acta Electronica Sinica, 2025, 53(10): 3551-3565. DOI：10.12263/DZXB.20250307

摘要

在城市环境下，人、车等对象的移动轨迹数据蕴含着丰富的居民活动信息，对城市规划、交通管控和流行病传播分析等具有重要价值.然而，隐私保护和商业机密等因素极大地限制了轨迹数据的共享和使用.生成能够保留真实轨迹特征的合成轨迹，以代替真实轨迹发布应用，已成为突破此限制的一种优选方案.近期，基于深度学习技术的轨迹生成研究颇受学术和工业界的关注，基于生成对抗网络（Generative Adversarial Networks，GAN）、扩散模型等的轨迹模型相继提出.但现有轨迹生成模型存在两大局限：其一，未能有效捕获人类移动轨迹中的全局空间依赖关系；其二，未能有效建模城市环境对轨迹生成的影

响，导致生成的轨迹与真实场景存在偏差.为此，本文提出融合多源城市环境信息的知识图谱驱动轨迹生成模型（urban Trajectory Generation via Knowledge Graph-enhanced multi-source context fusion，KG-TrajGen）.首先，该模型整合路网拓扑数据、兴趣点（Point Of Interest，POI）、功能区域划分情况等关键的多源城市环境数据，分别构建基础的道路知识图谱（Road Knowledge Graph，RKG）和环境语义增强型道路知识图谱（Environment-semantics-enhanced Road Knowledge Graph，E-RKG），并采用关系图卷积网络（Relational Graph Convolutional Network，R-GCN），从RKG中学习路段基础嵌入，以同时捕捉道路间的局部和全局空间依赖关系，采用结构感知的知识图谱嵌入方法，从E-RKG中学习城市环境知识，赋予模型环境感知能力，以进一步丰富路段嵌入特征.其次，采用Transformer解码器模型，从历史轨迹数据中学习城市中的人类活动模式特征，获取到历史轨迹数据增强的路段嵌入特征.最后，通过有效融合知识图谱增强的路段嵌入和历史轨迹数据增强的路段嵌入特征，以自回归方式实现环境感知的细粒度轨迹生成.在两个开源的真实轨迹数据集上的实验表明：KG-TrajGen在统计特征误差、频繁模式特征误差和轨迹误差方面的指标显著优于基线方法，且生成的轨迹能够在交通流量预测这一下游轨迹分析任务上也优于基线方法，充分验证了KG-TrajGen模型的有效性.KG-TrajGen模型的代码可在

https：//github.com/trajgen/KG-TrajGen

https://github.com/trajgen/KG-TrajGen

获得.

Abstract

Mobility trajectory data of individuals

vehicles

and other objects in urban environments contains rich information about residents’ activities

which is highly valuable for urban planning

traffic management

and epidemic spread analysis. However

privacy protection and commercial confidentiality significantly restrict the sharing and utilization of trajectory data. Generating synthetic trajectories that preserve the characteristics of real trajectories to replace real ones for release and application has become a preferred solution to overcome these limitations. Recently

deep learning-based trajectory generation research has attracted considerable attention from both academia and industry

with various trajectory models based on generative adversarial networks

diffusion models

and others being successively proposed. Nevertheless

existing trajectory generation models suffer from two major limitations: first

they fail to effectively capture global spatial dependencies in human mobility patterns; second

they inadequately model the influence of urban environments on trajectory generation

leading to deviations between generated trajectories and real-world scenarios. To address this

this paper proposes a knowledge graph-driven trajectory generation model integrating multi-source urban environmental information

named urban trajectory generation via knowledge graph-enhanced multi-source context fusion (KG-TrajGen). The model integrates key multi-source urban environmental data

including road network topology

points of interest (POI)

and functional zone classifications

to construct a foundational road knowledge graph (RKG) and an environment-semantics-enhanced road knowledge graph (E-RKG). A relational graph convolutional network is employed to learn basic road segment embeddings from the RKG

simultaneously capturing both local and global spatial dependencies among roads. Additionally

a structure-aware knowledge graph embedding method is used to extract urban environmental knowledge from the E-RKG

endowing the model with environmental awareness and enriching the road segment embedding features. Subsequently

a Transformer decoder model learns human activity pattern features from historical trajectory data to obtain trajectory history-enhanced road segment embeddings. Finally

by effectively fusing the knowledge graph-enhanced and historical trajectory-enhanced road segment embeddings

the model generates environment-aware

fine-grained trajectories in an autoregressive manner. Experiments on two open source real world trajectory datasets demonstrate that KG-TrajGen significantly outperforms baseline methods in terms of statistical feature error

frequent pattern feature error

and trajectory error metrics. Moreover

the generated trajectories perform better than those from baseline methods in downstream trajectory analysis tasks such as traffic flow prediction

fully validating the effectiveness of the KG-TrajGen model. The code for KG-TrajGen is available at

https://github.com/trajgen/KG-TrajGen

关键词

Keywords

references

JIANG W J , ZHAO W X , WANG J Y , et al . Continuous trajectory generation based on two-stage GAN [C ] // Proceedings of the AAAI Conference on Artificial Intelligence . Washington : AAAI , 2023 , 37 ( 4 ): 4374 - 4382 .

FENG J , YANG Z Y , XU F L , et al . Learning to simulate human mobility [C ] // Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining . New York : ACM , 2020 : 3426 - 3433 .

DU Y T , HU Y J , ZHANG Z K , et al . LDPTrace: Locally differentially private trajectory synthesis [J ] . Proceedings of the VLDB Endowment , 2023 , 16 ( 8 ): 1897 - 1909 .

曾卓 , 汪成亮 , 马飞 . 基于差分隐私的活动模式保护与时空轨迹发布方法 [J ] . 电子学报 , 2023 , 51 ( 3 ): 552 - 563 .

ZENG Z , WANG C L , MA F . Differentially private activity pattern and spatial-temporal trajectory publication [J ] . Acta Electronica Sinica , 2023 , 51 ( 3 ): 552 - 563 . (in Chinese)

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)

SIMINI F , BARLACCHI G , LUCA M , et al . A Deep Gravity model for mobility flows generation [J ] . Nature Communications , 2021 , 12 : 6576 .

JIANG S , YANG Y X , GUPTA S , et al . The TimeGeo modeling framework for urban motility without travel surveys [J ] . PNAS , 2016 , 113 ( 37 ): E5370 - E5378 .

CAO C , LI M . Generating mobility trajectories with retained data utility [C ] // Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining . New York : ACM , 2021 : 2610 - 2620 .

WANG X R , LIU X Y , LU Z T , et al . Large scale GPS trajectory generation using map based on two stage GAN [J ] . Journal of Data Science , 2021 , 19 ( 1 ): 126 - 141 .

ZHU Q Y , CHEN Y Z , WANG H , et al . A knowledge-enhanced framework for imitative transportation trajectory generation [C ] // 2022 IEEE International Conference on Data Mining . Piscataway : IEEE , 2023 : 823 - 832 .

WANG Y , CAO J , HUANG W J , et al . Spatiotemporal gated traffic trajectory simulation with semantic-aware graph learning [J ] . Information Fusion , 2024 , 108 : 102404 .

NING Y , LIU H , WANG H , et al . UUKG: Unified urban knowledge graph dataset for urban spatiotemporal prediction [C ] // Advances in Neural Information Processing Systems 36 . San Diego : NeurIPS , 2023 : 62442 - 62456 .

CAO Z S , XU Q Q , YANG Z Y , et al . Geometry interaction knowledge graph embeddings [J ] . Proceedings of the AAAI Conference on Artificial Intelligence , 2022 , 36 ( 5 ): 5521 - 5529 .

LIN Y , WAN H Y , GUO S N , et al . Pre-training context and time aware location embeddings from spatial-temporal trajectories for user next location prediction [J ] . Proceedings of the AAAI Conference on Artificial Intelligence , 2021 , 35 ( 5 ): 4241 - 4248 .

WANG H D , YU Q H , LIU Y , et al . Spatio-temporal urban knowledge graph enabled mobility prediction [J ] . Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies , 2021 , 5 ( 4 ): 1 - 24 .

CHEN L , YING Y K , LYU D D , et al . A multi-task embedding based personalized POI recommendation method [J ] . CCF Transactions on Pervasive Computing and Interaction , 2021 , 3 ( 3 ): 253 - 269 .

ZHU Y S , YE Y C , ZHANG S Y , et al . DiffTraj: Generating GPS trajectory with diffusion probabilistic model [C ] // The 37th Conference on Neural Information Processing Systems . New Orleans : NeurIPS Foundation , 2023 : 65168 - 65188 .

HAYDARI A , CHEN D J , LAI Z F , et al . MobilityGPT: Enhanced human mobility modeling with a GPT model [EB/OL ] . ( 2024-05-23 )[ 2025-10-06 ] . https://arXiv.org/abs/2402.03264 https://arXiv.org/abs/2402.03264 .

MAGLARAS L A , KATSAROS D . Social clustering of vehicles based on semi-Markov processes [J ] . IEEE Transactions on Vehicular Technology , 2016 , 65 ( 1 ): 318 - 332 .

YU L T , ZHANG W N , WANG J , et al . SeqGAN: Sequence generative adversarial nets with policy gradient [J ] . Proceedings of the AAAI Conference on Artificial Intelligence , 2017 , 31 ( 1 ): 2852 - 2858 .

FARHA Y ABU , RICHARD A , GALL J . When will you do what? - Anticipating temporal occurrences of activities [C ] // 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition . Piscataway : IEEE , 2018 : 5343 - 5352 .

HUANG D , SONG X , FAN Z P , et al . A variational autoencoder based generative model of urban human mobility [C ] // 2019 IEEE Conference on Multimedia Information Processing and Retrieval . Piscataway : IEEE , 2019 : 425 - 430 .

RAO J , GAO S , KANG Y , et al . LSTM-TrajGAN: A deep learning approach to trajectory privacy protection [EB/OL ] . ( 2020-06-14 )[ 2025-10-06 ] . https://arxiv.org/abs/2006.10521 https://arxiv.org/abs/2006.10521 .

ZHANG L M , ZHAO L , PFOSER D . Factorized deep generative models for end-to-end trajectory generation with spatiotemporal validity constraints [C ] // Proceedings of the 30th International Conference on Advances in Geographic Information Systems . New York : ACM , 2022 : 1 - 12 .

ZHU Y S , YU J J , ZHAO X Y , et al . ControlTraj: Controllable trajectory generation with topology-constrained diffusion model [C ] // Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining . New York : ACM , 2024 : 4676 - 4687 .

SONG Y W , DING J T , YUAN J , et al . Controllable human trajectory generation using profile-guided latent diffusion [J ] . ACM Transactions on Knowledge Discovery from Data , 2024 , 19 ( 1 ): 1 - 25 .

SHAO C Y , XU F L , FAN B B , et al . Chain-of-planned-behaviour workflow elicits few-shot mobility generation in LLMs [EB/OL ] . ( 2024-06-05 )[ 2025-10-16 ] . https://arXiv.org/abs/2402.09836 https://arXiv.org/abs/2402.09836 .

CHUNG J , GULCEHRE C , CHO K , et al . Empirical evaluation of gated recurrent neural networks on sequence modeling [EB/OL ] . ( 2014-12-11 )[ 2025-10-16 ] . https://arXiv.org/abs/1412.3555 https://arXiv.org/abs/1412.3555 .

ZIAT A , DELASALLES E , DENOYER L , et al . Spatio-temporal neural networks for space-time series forecasting and relations discovery [C ] // 2017 IEEE International Conference on Data Mining . Piscataway : IEEE , 2017 : 705 - 714 .

BAI L , YAO L N , LI C , et al . Adaptive graph convolutional recurrent network for traffic forecasting [C ] // Proceedings of the 34th International Conference on Neural Information Processing Systems . New York : ACM , 2020 : 17804 - 17815 .

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Driving Behavior Recognition Based on Multimodal Physiological Feature Fusion

Large Language Model-Assisted Cloud-Edge Collaborative Workflow Scheduling Algorithm

A Medical Image Segmentation Network Based on Cross-Visual State Space and Multi-Branch Interactive Attention

Differentially Private with Sparse and Smooth Self-Distillation

STUDY OF THE DEGRADATION CHARACTERISTICS FOR THE InGaAsP/InP DH LEDs

Related Author

SHAO Shu-yu

ZHANG Yang

YAN Wen-jing

LI Guang-rong

LI Guang-jun

SHANG Jing

WU Wen-tai

WANG Ze-ping

Related Institution

School of Intelligent Engineering and Supply Chain Innovation, Beijing Wuzi University

Beijing Key Laboratory of Intelligent Logistics System

School of Computer and Artificial Intelligence, Beijing Technology and Business University

College of Information Science and Technology, Jinan University

China Mobile Information Technology Co., Ltd.

⁰