张海明,徐胜华,陈煦,虞赢赢,程语然

• 1:辽宁工程技术大学测绘与地理科学学院
• 2:中国测绘科学研究院
• 3:中国国土勘测规划院

摘要(Abstract)：

针对下一个兴趣点推荐中异构语义表达不充分、模态不匹配及意图挖掘受限等问题，提出一种大语言模型时空语义特征对齐推理的推荐方法。通过门控融合策略构建兴趣点统一语义向量，利用滑动窗口与时间间隔编码捕捉局部移动模式；采用前缀生成网络将轨迹片段映射为虚拟词元序列，并通过层级注入策略实现跨模态语义对齐；结合包含用户画像与任务指令的结构化提示引导模型推理，在固定大语言模型主要参数下进行预测。在Foursquare和微博数据集上的实验表明，该方法在首项准确率和平均倒数排名上显著优于主流基线模型，可应用于个性化位置推荐和商业布局优化等领域。

关键词(KeyWords)： 兴趣点;推荐系统;时序行为;大语言模型;基于位置的社交网络

Abstract:

Keywords:

基金项目(Foundation): 国家自然科学基金项目(42371478)

作者(Author): 张海明,徐胜华,陈煦,虞赢赢,程语然

DOI: 10.16251/j.cnki.1009-2307.2026.11.015

参考文献(References)：

• [1] GAO K,YANG X,WU C,et al.Exploiting location-based context for POI recommendation when traveling to a new region[J].IEEE Access,2020,8:52404-52412.
• [2] 王勇，马钰，徐胜华，等.兴趣点推荐方法研究进展与展望[J].测绘科学，2023,48(12):217-224.(WANG Yong,MA Yu,XU Shenghua,et al.Research progress and prospects of point of interest recommendation methods[J].Science of Surveying and Mapping,2023,48(12):217-224.)
• [3] RENDLE S,FREUDENTHALER C,SCHMIDT-THIEME L.Factorizing personalized Markov chains for next-basket recommendation[C]//Proceedings of the 19th International Conference on World Wide Web.[S.l.]:[s.n.],2010:811-820.
• [4] ZHAO P,LUO A,LIU Y,et al.Where to go next:a spatio-temporal gated network for next POI recommendation[J].IEEE Transactions on Knowledge and Data Engineering,2020,34(5):2512-2524.
• [5] XU H,DING W,SHEN W,et al.Deep convolutional recurrent model for region recommendation with spatial and temporal contexts[J].Ad Hoc Networks,2022,129:102545.
• [6] 柴瑞敏，殷臣，孟祥福，等.基于时空循环神经网络的下一个兴趣点推荐方法[J].智能系统学报，2021,16(3):407-415.(CHAI Ruimin,YIN Chen,MENG Xiangfu,et al.A recurrent neural network model based on spatial and temporal information for the next point of interest recommendation[J].CAAI Transactions on Intelligent Systems,2021,16(3):407-415.)
• [7] WU Y,LI K,ZHAO G,et al.Personalized long-and short-term preference learning for next POI recommendation[J].IEEE Transactions on Knowledge and Data Engineering,2020,34(4):1944-1957.
• [8] 何璇，徐胜华，车向红，等.顾及周期性时序行为与社交关系的下一个兴趣点推荐[J].测绘通报，2025 (10):106-113.(HE Xuan,XU Shenghua,CHE Xianghong,et al.Considering periodic temporal behaviors and social relationships for next point-of-interest recommendation[J].Bulletin of Surveying and Mapping,2025 (10):106-113.)
• [9] LUO Y,LIU Q,LIU Z.STAN:spatio-temporal attention network for next location recommendation[C]//International World Wide Web Conference(IW3C2),Ljubljana,Slovenia.New York:ACM,2021:2177-2185.
• [10] LI X,HU R,WANG Z.Beyond fixed time and space:next POI recommendation via multi-grained context and correlation[J].Neural Computing and Applications,2023,35(1):907-920.
• [11] 刘广腾，王峰，吴中博.下一个兴趣点推荐算法综述[J].计算机科学与探索，2025,19(7):1747-1770.(LIU Guangteng,WANG Feng,WU Zhongbo.Survey on next point-of-interest recommendation algorithms[J].Journal of Frontiers of Computer Science & Technology,2025,19(7):1747-1770.)
• [12] YANG S,LIU J,ZHAO K.GETNext:trajectory flow map enhanced transformer for next POI recommendation[C]//Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval.New York:ACM,2022:1144-1153.
• [13] YU C,SHI L,ZHAO Y.Trajectory-and friendship-aware graph neural network with Transformer for next POI recommendation[J].ISPRS International Journal of Geo-Information,2025,14(5):192.
• [14] 李鹏飞，贺洋，毋建宏.融合全局特征的时空网络兴趣点推荐算法[J].计算机工程与应用，2024,60(11):75-83.(LI Pengfei,HE Yang,WU Jianhong.Spatio-temporal network interest point recommendation algorithm fusing global features[J].Computer Engineering and Applications,2024,60(11):75-83.)
• [15] 彭瑾，史艳翠，刘凌云.基于双粒度序列融合的下一个兴趣点推荐[J].计算机系统应用，2025,34(1):128-136.(PENG Jin,SHI Yancui,LIU Lingyun.Next point-of-interest recommendation based on dual-granularity sequence fusion[J].Computer Systems & Applications,2025,34(1):128-136.)
• [16] 李婉秋，张超群，汤卫东，等.面向下一个兴趣点推荐的细粒度时空多语义超图学习[J].计算机应用研究，2025,42(2):398-405.(LI Wanqiu,ZHANG Chaoqun,TANG Weidong,et al.Multi-semantic hypergraph learning with fine-grained spatio-temporal information for next POI recommendation[J].Application Research of Computers,2025,42(2):398-405.)
• [17] WANG X,FUKUMOTO F,LI J,et al.STaTRL:Spatial-temporal and text representation learning for POI recommendation[J].Applied Intelligence,2023,53(7):8286-8301.
• [18] 郭旦怀，张鸣珂，贾楠，等.融合深度学习技术的用户兴趣点推荐研究综述[J].武汉大学学报(信息科学版),2020,45(12):1890-1902.(GUO Danhuai,ZHANG Mingke,JIA Nan,et al.Survey of point-of-interest recommendation research fused with deep learning[J].Geomatics and Information Science of Wuhan University,2020,45(12):1890-1902.)
• [19] 迟晋浙，刘纪平，徐胜华，等.自适应门控机制嵌入图神经网络的下一个POI推荐[J].测绘通报，2025 (7):90-96.(CHI Jinzhe,LIU Jiping,XU Shenghua,et al.Next POI recommendation with an adaptive gating mechanism embedded in graph neural networks[J].Bulletin of Surveying and Mapping,2025 (7):90-96.)
• [20] WU Y,LI K,ZHAO G,et al.Long-and short-term preference learning for next POI recommendation[C]//Proceedings of the 28th ACM International Conference on Information and Knowledge Management.New York:ACM,2019:2301-2304.
• [21] HUANG T,PAN X,CAI X,et al.Learning time slot preferences via mobility tree for next POI recommendation[C]//Proceedings of the 38th AAAI Conference on Artificial Intelligence,Vancouver,Canada.Palo Alto,CA,USA:AAAI Press,2024,38(8):8535-8543.