第16卷第3期 智能系统学报 Vol.16 No.3 2021年5月 CAAI Transactions on Intelligent Systems May 2021 D0L:10.11992tis.202004009 基于时空循环神经网络的下一个兴趣点推荐方法 柴瑞敏，殷臣，孟祥福，张霄雁，关昕，齐雪月 (辽宁工程技术大学电子与信息工程学院，辽宁葫芦岛125105)】 摘要：下一个兴趣点推荐已经成为基于位置的社交网络(location--based social networks,,LBSNs)中一个重要任 务。现有的模型没有深入考虑相邻签到兴趣点之间的转移时空信息，无法对用户访问下一个兴趣点的长短时 间偏好和远近距离偏好进行有效建模。本文通过对循环神经网络(recurrent neural network,RNN)进行扩展，提 出一个新的基于会话的时空循环t神经网络模型(sesson-based spatial--temporal recurrent neural network,SST RNN)用于下一个兴趣点推荐。该模型通过设置时间转移矩阵和空间转移矩阵分别对用户的时间和空间偏好 信息进行建模，综合考虑连续签到兴趣点的序列信息、时空信息以及用户偏好进行下一个兴趣点推荐。通过 在2个真实公开的数据集上进行实验，结果显示本文提出的SST-RNN模型的推荐效果比主流的推荐模型有显 著提升。在Foursquare和CA数据集上，ACC@5评价指标分别提升了36.38%和13.81%，MAP评价指标分别提 升了30.72%和17.26%。 关键词：下一个兴趣点推荐：基于位置的社交网络：循环神经网络；序列信息：时间偏好：空间偏好：用户偏好： 会话 中图分类号：TP183文献标志码：A文章编号：1673-4785(2021)03-0407-09 中文引用格式：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，智能系统学报，2021,16(3)： 407-415. 英文引用格式：CHAI Ruimin,YIN Chen,MENG Xiangfu,etal.A recurrent neural network model based on spatial and temporal information for the next point of interest recommendation J.CAAl transactions on intelligent systems,2021,16(3):407-415. A recurrent neural network model based on spatial and temporal information for the next point of interest recommendation CHAI Ruimin,YIN Chen,MENG Xiangfu,ZHANG Xiaoyan,GUAN Xin,QI Xueyue (School of Electronic and Information Engineering,Liaoning Technical University,Huludao 125105,China) Abstract:The next point-of-interest (POl)recommendation has become an important task in location-based social net- works.The existing models lack in-depth research on the temporal and spatial information transition between adjacent check-in POIs and cannot effectively model the long/short time and distance preferences of the users accessing the next POI.In response,this paper proposes a new session-based spatial-temporal recurrent neural network(SST-RNN)model that is used to recommend the next POI.This model takes advantage of the spatial transition matrix and temporal trans- ition matrix to respectively model the user's spatial and temporal preferences,and comprehensively considers the se- quence information and spatial-temporal information of consecutive check-in POIs as well as user preferences to do the next POI recommendation.Experimental results in two real open datasets show that the performance of the proposed SST-RNN model is significantly enhanced compared with the state-of-the-art models.On the Foursquare and CA data- sets,the ACC@5 is increased by 36.38%and 13.81%,and the MAP is increased by 30.72%and 17.26%,respectively. Keywords:next point of interest recommendation;location-based social networks,recurrent neural network;sequence information;temporal preferences;spatial preferences;user preferences;session 收稿日期：2020-04-09. 基金项目：国家自然科学基金面上项目(61772249) 随着基于位置的社交网络发展，一些基于位 通信作者：孟祥福.E-mail:marxi(@I26.com. 置服务的社交软件被广泛使用，如Foursquare、DOI: 10.11992/tis.202004009 基于时空循环神经网络的下一个兴趣点推荐方法 柴瑞敏，殷臣，孟祥福，张霄雁，关昕，齐雪月 （辽宁工程技术大学 电子与信息工程学院，辽宁 葫芦岛 125105） 摘 要：下一个兴趣点推荐已经成为基于位置的社交网络 (location-based social networks，LBSNs) 中一个重要任 务。现有的模型没有深入考虑相邻签到兴趣点之间的转移时空信息，无法对用户访问下一个兴趣点的长短时 间偏好和远近距离偏好进行有效建模。本文通过对循环神经网络 (recurrent neural network, RNN) 进行扩展，提 出一个新的基于会话的时空循环神经网络模型 (sesson-based spatial-temporal recurrent neural network, SST￾RNN) 用于下一个兴趣点推荐。该模型通过设置时间转移矩阵和空间转移矩阵分别对用户的时间和空间偏好 信息进行建模，综合考虑连续签到兴趣点的序列信息、时空信息以及用户偏好进行下一个兴趣点推荐。通过 在 2 个真实公开的数据集上进行实验，结果显示本文提出的 SST-RNN 模型的推荐效果比主流的推荐模型有显 著提升。在 Foursquare 和 CA 数据集上，ACC@5 评价指标分别提升了 36.38% 和 13.81%，MAP 评价指标分别提 升了 30.72% 和 17.26%。 关键词：下一个兴趣点推荐；基于位置的社交网络；循环神经网络；序列信息；时间偏好；空间偏好；用户偏好； 会话 中图分类号：TP183 文献标志码：A 文章编号：1673−4785(2021)03−0407−09 中文引用格式：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. 英文引用格式：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. A recurrent neural network model based on spatial and temporal information for the next point of interest recommendation CHAI Ruimin，YIN Chen，MENG Xiangfu，ZHANG Xiaoyan，GUAN Xin，QI Xueyue (School of Electronic and Information Engineering, Liaoning Technical University, Huludao 125105, China) Abstract: The next point-of-interest (POI) recommendation has become an important task in location-based social net￾works. The existing models lack in-depth research on the temporal and spatial information transition between adjacent check-in POIs and cannot effectively model the long/short time and distance preferences of the users accessing the next POI. In response, this paper proposes a new session-based spatial–temporal recurrent neural network (SST-RNN) model that is used to recommend the next POI. This model takes advantage of the spatial transition matrix and temporal trans￾ition matrix to respectively model the user’s spatial and temporal preferences, and comprehensively considers the se￾quence information and spatial–temporal information of consecutive check-in POIs as well as user preferences to do the next POI recommendation. Experimental results in two real open datasets show that the performance of the proposed SST-RNN model is significantly enhanced compared with the state-of-the-art models. On the Foursquare and CA data￾sets, the ACC@5 is increased by 36.38% and 13.81%, and the MAP is increased by 30.72% and 17.26%, respectively. Keywords: next point of interest recommendation; location-based social networks; recurrent neural network; sequence information; temporal preferences; spatial preferences; user preferences; session 随着基于位置的社交网络发展，一些基于位 置服务的社交软件被广泛使用，如 Foursquare、 收稿日期：2020−04−09. 基金项目：国家自然科学基金面上项目 (61772249). 通信作者：孟祥福. E-mail：marxi@126.com. 第 16 卷第 3 期 智 能 系 统 学 报 Vol.16 No.3 2021 年 5 月 CAAI Transactions on Intelligent Systems May 2021