人工智能大数据技术在治理虚拟货币刑事犯罪中的应用研究

打开文本图片集

中图分类号：TN915.08-34；TP39 文献标识码：A 文章编号：1004-373X（2025）16-0167-05

Research onapplicationof artificial intelligenceandbigdata technology in governance of virtual currency criminal offense

LIUHuan²，XIAOWei³ （1.Xi'anJiaotongUniversity，Xi'an710o49，China; 2.Shaanxi Public SecurityNew NetworkFinancial CrimeResearch Center，ShaanxiPoliceColege，Xi’an7Oo21，China; 3.ShaanxiProvincial Public SecurityDepartment，Xi'an71oo18，China）

Abstract：In allusion to thegovernance needs of virtual currencycriminal offense，anenhanced detection modelbasedon relationalgraphconvolutionalnetwork（RGCN）isproposed.Inthemodel，originalfeatureandcross-temporalfeature concatenationmechanismareintegratedinovativelyintothegraphembeddnglayerdesign，efectivelymitigatingtheinfoation losscausedbytheredundantconvolution.Therandomforestensemblelearningstrategyisusedtooptimizethefeature representationinthegraphembeddinglayer，significantlyimprovingmodelrecognitionaccuracy.Theconstructedthre-layer RGCN framework canreducecomputationaloverheadbymeansofspatio-temporalfeaturefusion strategyandoptimizetherecal rate.TheexperimentalresultsdemonstratethattheLSTM-basedRGCNcanperformthebestintherecallindex，whichis67.6% and is 4.4% higher than the GRU variant;while the GRU structure focuses moreon the optimization of accuracy，with an accuracy of 87.3% .Combining thecurrenturgent requirement for monitoring high-risk individuals inanti-money laundering scenarios，itdemonstratesthattheproposedmodelcanefectivelysupportvirtualcurencytransactionsupervisionpracticesby enhancing recall performance.

Keywords：artificial intelligence;virtual currency；criminal offense；dynamicgraph convolutional network；LSTM;GRU; recall optimization

0 引言

流通的媒介，以及作为在特定虚拟环境中流通的数字媒介，不仅支持服务与商品的便捷交换1，还具有独特的交易特性。（剩余7240字）