基于上下文全局空间图的轨迹用户链接  

Trajectory-user linking based on contextual global spatial graph

作　　者：侯萱 梁志贞[1,2] 张磊[1,2] 刘佰龙[1,2] 张雪飞[3] HOU Xuan;LIANG Zhizhen;ZHANG Lei;LIU Bailong;ZHANG Xuefei(Engineering Research Center of Mine Digitalization of the Ministry of Education,China University of Mining and Technology,Xuzhou 221116;School of Computer Science&Technology,China University of Mining and Technology,Xuzhou 221116;Jiangsu Hengwang Digital Technology Co.,Ltd.,Suzhou 215000,China)

机构地区：[1]中国矿业大学矿山数字化教育部工程研究中心,江苏徐州221116 [2]中国矿业大学计算机科学与技术学院,江苏徐州221116 [3]江苏恒旺数字科技有限责任公司,江苏苏州215000

出　　处：《计算机工程与科学》2025年第2期336-348,共13页Computer Engineering & Science

基　　金：中国矿业大学建设双一级专项资金(18ZZCX14)。

摘　　要：轨迹用户链接TUL是指判定目标轨迹所属用户,已成为一项重要的轨迹数据挖掘任务。尽管基于深度学习的模型在TUL研究中取得显著进展,但现有模型主要关注单个轨迹点的基本时空特征,忽略全局位置空间相关性、上下文信息和用户的多周期移动规律,导致TUL结果准确度不高。提出了一种基于上下文全局空间图的轨迹用户链接模型CGSG-TUL。在位置嵌入方面,根据历史轨迹构建上下文全局空间图,融入所有位置的邻近关系和类别等上下文信息,对位置的空间相关性有效建模。在时间编码方面,根据不同时间尺度对签入的时间戳进行编码,捕获用户的多周期移动规律。在Foursquare-NYK和Foursquare-TKY这两个真实数据集上的实验结果表明,CGSG-TUL性能比目前最好的基准模型GNNTUL的ACC@1和Marco-F 1分别平均提高2.50%和2.72%。Trajectory-user linking(TUL)refers to determining the user to whom a target trajectory belongs and has become an important trajectory data mining task.Although deep learning-based models have made significant progress in TUL research,existing approaches mainly focus on the basic spatiotemporal features of individual trajectory points,neglecting the global spatial correlation,contextual information,and users multi-periodic movement patterns,resulting in low accuracy in TUL results.In this regard,a trajectory-user linking model based on contextual global spatial graph(CGSG-TUL)is proposed.In terms of location embedding,a contextual global spatial graph is constructed based on historical trajectories,incorporating contextual information such as proximity relationships and categories of all locations.This effectively models the spatial correlations of locations.Regarding time encoding,the timestamps of check-ins are encoded according to different time scales to capture users multi-periodic movement patterns.Experimental results on two real datasets,Foursquare-NYK and Foursquare-TKY,demonstrate that CGSG-TUL outperforms the state-of-the-art baseline model GNNTUL,with an average improvement of 2.50%and 2.72%in terms of ACC@1 and Macro-F 1.

关 键 词：轨迹用户链接 上下文全局空间图 多周期移动规律 图神经网络 TRANSFORMER 

分 类 号：TP393.09[自动化与计算机技术—计算机应用技术]