| 349 | 0 | 1190 |
| 下载次数 | 被引频次 | 阅读次数 |
城市犯罪是危害居民生命和财产安全的重大社会问题，基于深度学习的已有犯罪预测研究鲜有考虑城市人口动态，难以实现精准预测。该文提出一种结合注意力机制的图卷积门控循环单元犯罪预测模型，同时考虑犯罪数据的近邻趋势特征、长期趋势特征及出租车流特征，并通过改进的时空特征通道注意力实现多特征数据的有效融合；在社区空间拓扑的基础上使用POI分布和社区间的出租车流共同表示社区间的相关关系，进而更全面地捕获犯罪在空间上的相互影响。基于美国芝加哥市犯罪数据集的实验结果表明，模型在盗窃犯罪预测的各项评价指标上均优于其他基线模型，且加入长期趋势特征和出租车流特征对盗窃犯罪预测结果提升显著；模型对伤害、刑事损坏及欺诈类犯罪的预测效果均较好，表明具备一定的泛化能力。
Abstract：Urban crime is a significant social problem that seriously endangers the lives and property of residents.Existing deep learning-based crime prediction studies rarely consider dynamic urban population changes,leading to inaccurate predictions.To address this,we propose an attention-graph convolutional gated recurrent unit(A-GCGRU)model.The model simultaneously incorporates neighboring trend features,long-term trend features,and taxi flow features to extract spatiotemporal dependencies,and employs an improved spatiotemporal channel attention mechanism to achieve effective fusion of multi-source features.Based on community spatial topology,the model represents inter-community correlations using POI(points of interest)and inter-community taxi origin-destination(OD)flows,thereby capturing the spatial interactions of crimes more comprehensively.Experimental results on the Chicago crime dataset demonstrate that A-GCGRU outperforms other baseline models in theft crime prediction across various evaluation metrics.The inclusion of long-term trend and taxi flow features significantly improves theft crime prediction accuracy.The model also exhibits superior predictive performance for battery,criminal damage,and deceptive practice,verifying its generalizability.
[1] BECONYTE G,GRUZAS K,SPIRIAJEVAS E.Areas of crime in cities:case study of Lithuania[J].ISPRS International J ournal of Geo-Information,2023,13(1):1.
[2]TAYEBI M A,ESTER M,GL?SSER U,et al.Crime tracer:activity space based crime location prediction[C]//2014 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining(ASONAM 2014).Beijing,China,2014.
[3]EGBERT S,LEESE M.Criminal futures:predictive policing and everyday police work[M].London:Routledge,2020.
[4] MALIK A,MACIEJEWSKI R,TOWERS S,et al.Proactive spatiotemporal resource allocation and predictive visual analytics for community policing and law enforcement[J].IEEE Transactions on Visualization and Computer Graphics,2014, 20(12):1863-1872.
[5]MOHLER G O,SHORT M B,BRANTINGHAM P J,et al Selfexciting point process modeling of crime[J].Journal of theAmerican Statistical Association,2011,106(493):100-108.
[6] WORTLEY R K,TOWNSLEY M.Environmental criminology and crime analysis:situating the theory,analytic approach and application[M]//WORTLEY R K,TOWNSLEY M.Environmental Criminology and Crime Analysis.London:Routledge,2016.
[7]KENNEDY L W,CAPLAN J M,PIZA E,et al.Risk clusters,hotspots, and spatial intelligence:risk terrain modeling as an algorithm for police resource allocation strategies[J].Journal of Quantitative Criminology,2011,27:339-362.
[8]柳林,纪佳楷,宋广文,等.基于犯罪空间分异和建成环境的公共场所侵财犯罪热点预测[J].地球信息科学学报,2019,21(11):1655-1668.
[9]赵丹,杜萍,刘涛,等.融合图自编码器与GRU的城市盗窃犯罪时空分布预测模型[J].地球信息科学学报,2023,25(7):1448-1463.
[10]BENDLER J,RATKU A,NEUMANN D.Crime mapping through geo-spatial social media activity[C]//Thirty Fifth International Conference on Information Systems.Auckland:Association for Information Systems,2014.
[11]LAW J,QUICK M,CHAN P W.Analyzing hotspots of crime using a Bayesian spatiotemporal modeling approach:a case study of violent crime in the greater Toronto area[J].Geographical Analysis,2015,47(1):1-19.
[12]HAN X G,HU X F,WU H G,et al.Risk prediction of theft crimes in urban communities:an integrated model of LSTM and ST-GCN[J].IEEE Access,2020,8:217222-217230.
[13]SONG G W,LIU L,BERNASCO W,et al.Testing indicators of risk populations for theft from the person across space and time:the significance of mobility and outdoor activity[J].Annals of the American Association of Geographers,2018,108(5):1370-1388.
[14]WANG H J,YAO H X,KIFER D,et al.Non-stationary model for crime rate inference using modern urban data[J].IEEE Transactions on Big Data,2019,5(2):180-194.
[15]VOMFELL L,H?RDLE W K,LESSMANN S.Improving crime count forecasts using Twitter and taxi data[J].Decision Support Systems,2018,113:73-85.
[16] GERBER M S.Predicting crime using Twitter and kernel density estimation[J].Decision Support Systems,2014,61:115-125.
[17] KADAR C,PLETIKOSA I.Mining large-scale human mobility data for long-term crime prediction[J].EPJ Data Science,2018,7(1):1-27.
[18]SONG G W,BERNASCO W,LIU L,et al.Crime feeds on legal activities:daily mobility flows help to explain thieves'target location choices[J].Journal of Quantitative Criminology,2019,35(4):831-854.
[19]MACARTHUR R H,PIANKA E R.On optimal use of a patchy environment[J].The American Naturalist,1966,100(916):603-609.
[20]ROSSMO D K.Geographic profiling[M].New York:Routledge,1999.
[21]BRANTINGHAM P,BRANTINGHAM P.Crime pattern theory[M]//WORTLEY R,MAZEROLLE L.Environmental Criminology and Crime Analysis.London:Willan,2013:100-116.
[22]Chicago CompStat 2023 Crime Statistics[R/OL].Chicago Police Department,2024.[2024-10-25].https://home.chicagopolice.org/statistics-data/crime-statistics/.
[23]赵美风,张蓝青,汪德根.生活圈视角下城市公共服务设施配置分异及影响因素:基于京津冀地区不同规模城市比较[J].城市发展研究,2024,31(10):77-86.
[24]赵玉敏.“15分钟”视角下的城市居民区公共服务设施配置研究[D].武汉:武汉大学,2020.
[25]丁彦文,许捍卫,汪成昊.融合OSM路网与POI数据的城市功能区识别研究[J].地理与地理信息科学,2020,36(4):57-63.
[26]梅康,刘小勤,沐超,等.基于自适应指数加权移动平均滤波的快速去雾算法[J].中国激光,2020,47(1):250-259.
[27] BRUNA J,ZAREMBA W,SZLAM A,et al.Spectral networks and locally connected networks on graphs[EB/OL].(2013-12-21)[2024-03-10].https://arxiv.org/abs/1312.6203.
[28] LAWRENCE S,GILES C L, TSOI A C,et al.Face recognition:a convolutional neural-network approach[J].IEEE Transactions on Neural Networks,1997,8(1):98-113.
[29]陈喜群,周凌霄,曹震.基于图卷积网络的路网短时交通流预测研究[J].交通运输系统工程与信息,2020,20(4):49-55.
[30]邓培芳,徐科杰,黄鸿.基于CNN-GCN双流网络的高分辨率遥感影像场景分类[J].遥感学报,2021,25(11):2270-2282.
[31]CHO K,VAN MERRI?NBOER B,BAHDANAU D,et al.On the properties of neural machine translation:encoder-decoder approaches[EB/OL].(2014-09-03)[2025-01-10].https://arxiv.org/abs/1409.1259.
[32] PALM R B. Prediction as a candidate for learning deep hierarchical models of Lyngby:Technical University of Denmark,2012.
[33]王鑫,吴际,刘超,等.基于LSTM循环神经网络的故障时间序列预测[J].北京航空航天大学学报,2018,44(4):772-784.
[34]陈涛,葛小三.融合道路特征的动态卷积遥感影像道路提取网络模型[J].地理与地理信息科学,2025,41(2):23-30.
[35] WOO S,PARK J,LEE J Y,et al.CBAM:convolutional block attention module[C]//Computer Vision-ECCV 2018.Cham:Springer,2018:3-19.
[36]GIRSHICK R.Fast R-CNN[C]//2015 IEEE International Conference on Computer Vision(ICCV).Santiago:IEEE,2015:1440-1448.
[37] MAHFOUD M,BERNASCO W,BHULAI S,et al.Forecasting spatio-temporal variation in residential burglary with the integrated Laplace approximation framework:effects of crime generators,street networks,and prior crimes[J].Journal of Quantitative Criminology,2021,37(4):835-862.
[38] ZHANG Q,YUAN P M,ZHOU Q Y,et al.Mixed spatial-temporal characteristics based crime hot spots prediction[C]//2016IEEE 20th International Conference on Computer Supported Cooperative Work in Design(CSCWD).Nanchang:IEEE,2016:97-101.
[39] BAI JD,ZHU J W,SONG Y J,et al.A3T-GCN:attention temporal graph convolutional network for traffic forecasting[J].ISPRS International Journal of Geo-information,2021,10(7):485.
[40] 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.
[41] YU B,YIN H T,ZHU Z X.Spatio-temporal graph convolutional networks:a deep learning framework for traffic forecasting[EB/OL].(2017-09-14)[2024-03-11].https://arxiv.org/abs/1709.04875.
基本信息:
中图分类号:D917.6;U491;P208
引用信息:
[1]张子皓 ,杜萍 ,刘涛 ,等.结合出租车OD流特征的城市犯罪时空分布预测[J].地理与地理信息科学,2025,41(05):1-9.
基金信息:
国家自然科学基金项目“面向应急制图的非常规突发事件地理场景建模研究”（42061060）;国家自然科学基金项目“灾害场景下应急地图需求一体化建模”（42261076）; 甘肃省重大科技专项“公路资产数字化关键技术研究”（22ZD6GA010）; 国家资助博士后研究人员计划项目（GZC20231020）