MicrocomputerApplicationsVol.39,No.8,2023文章编号:1007-757X(2023)08-0013-03摘要:当前,医院信息系统(HIS)已成为医院信息化建设的重要内容,但HIS与财务数据库的接入仍然采用传统的方式,导致财务重要数据存在一定的安全隐患。为了有效消除用户异常行为对医院财务数据库所构成的安全隐患,设计一种财务数据库异常检测技术。通过调取财务数据库运行日志中的用户查询内容及相应结果,采用k-means聚类算法进行用户分组,采用NavieBayes算法构建异常检测模型。应用测试结果表明,与传统的用户行为轮廊算法相比,所提出的算法准确率提高了7.06个百分点,综合F1值提高了3.33个百分点,此外,在大幅度缩减计算量的基础上模型训练时间缩短了81%,极大地提高了财务数据的安全性。关键词:财务数据库;异常检测;NavieBayes算法;HIS;安全隐惠中图分类号:TP393.08(1.TheNuclearIndustryGeneralHospital(TheSecondAffiliatedHospitalofSoochowUniversity),Suzhou215000,China;2.TheSuzhouBranchofShanghaiPudongDevelopmentBank,Suzhou215028,China)Abstract:Currently,hospitalinformationsystem(HIS)hasbecomeanimportantpartofhospitalinformationconstruction.However,theaccessbetweenHISandfinancialdatabasestilladoptsthetraditionalway,resultingincertainsecurityrisks.Inordertoeffectivelyeliminatethesecurityrisksposedbyabnormaluserbehaviortohospitalfinancialdatabase,ananomalyde-tectiontechnologyisproposedanddesigned.Byretrievingtheuserquerycontentsandcorrespondingresultsintheoperationofthefinancialdatabase,thek-meansclusteringalgorithmisusedtogroupusers,andNavieBayesalgorithmisusedtobuildana-nomalydetectionmodel.Theapplicationtestresultsshowthatcomparedwiththetraditionaluserbehaviorcontouralgorithm,theaccuracyoftheproposedalgorithmisimprovedby7.06%,andthecomprehensiveFivalueisimprovedby3.33%.Inaddi-tion,onthebasisofgreatlyreducingtheamountofcalculation,themodeltrainingtimeisshortenedby81%,whichgreatlyimprovesthesecurityoffinancialdata.Keywords:financialdatabase;anomalydetection;NavieBayesalgorithm;hospitalinformationsystem(HIS);securityrisk0引言近年来,随着互联网用户的急剧增加,各种网络入侵事件层出不穷,入侵检测系统的研发随之加快。异常检测作为人侵检测的...