Page 48 - 2023年第54卷第6期
P. 48
[21] ZHAOXJ,JIAY,LIAP,etal.Multi - sourceknowledgefusion:asurvey[J].WorldWideWeb - Internetand
WebInformationSystems ,2020,23(4):2567 - 2592.
[22] ZOUL,ZSUM T.Graph - basedRDFdatamanagement[J].DataScienceandEngineering,2017,2:56 - 70.
[23] WYLOTM,HAUSWIRTH M,CUDRE - MAUROUXP,etal.RDFdatastorageandQueryProcessingSchemes:
Asurvey[J].ACM ComputingSurveys,2019,51(4):1 - 36.
[24] RAULF.Extractingcompanynamesfrom text[C]??Proceedingsofthe7thIEEEConferenceonArtificialIntelli
genceApplications,1991.
[25] LIUX,ZHANG S,WEIF,etal.Recognizingnamedentitiesintweets[C]??Proceedingsofthe49thAnnual
MeetingoftheAssociationforComputationalLinguistics ,2011.
[26] HUANGZH,XUW,YUK.BidirectionalLSTM- CRFmodelsforsequencetagging[EB?OL].(2015 - 08 - 09)
[ 2022 - 10 - 16].https:??arxiv.org?pdf?1508.01991v1.pdf.
[27] DEVLINJ,CHANGM W,LEEK,etal.BERT:Pretrainingofdeepbidirectionaltransformersforlanguageun
derstanding [EB?OL].(2019 - 05 - 24)[2022 - 10 - 17].https:??arxiv.org?pdf?1810.04805.pdf.
[28] 李晓庆,唐昊,司加胜,等.面向混合属性数据集的改进半监督 FCM 聚类方法[J].自动化学报,2018,
44(12):2259 - 2268.
[29] MIKOLOVT,CHENK,DEAN J,etal.Efficientestimationofwordrepresentationsinvectorspace[EB?OL].
(2013 - 09 - 27)[2022 - 10 - 17].https:??arxiv.org?pdf?1301.3781v3.pdf.
[30] 官赛萍,靳小龙,贾岩涛,等.面向知识图谱的知识推理研究进展[J].软件学报,2018(10):2966 - 2994.
[31] CHENY,GOLDBERGS,WANGDZ,etal.Ontologicalpathfinding:miningfirst - orderknowledgefrom large
knowledgebases [C]??Proceedingsof2016InternationalConferenceonManagementofData,2016.
Aknowledge - drivenapproachforintelligentgenerationofhydraulicengineeringcontingency
plans:AcasestudyoftheMiddleRouteofSouth - to - NorthWaterDiversionProject
1
1,2
3
4
LIUXuemei ,LUHankang,LIHairui,HUAIXianfeng,CHENXiaolu 4
(1.SchoolofInformationEngineering,NorthChinaUniversityofWaterResourcesandElectricPower,Zhengzhou 450000,China;
2.CollaborativeInnovationCenterforEfficientUtilizationofWaterResources,Zhengzhou 450000,China;
3.SchoolofManagementandEconomics,NorthChinaUniversityofWaterResourcesandElectricPower,Zhengzhou 450000,China;
4.ChinaSouth - to - NorthWaterDiversionGroupMiddleLineCo.,Ltd,Beijing 100038,China)
Abstract:Traditionalemergencyplansforhydraulicengineeringprojectshaveproblemssuchaslowdigitisation,
poorcontentrelevanceandinsufficientintelligentaidfordecision - making.Inthispaper,weusetheknowledge
graphanddeeplearningtechnologytocreateanintelligentgenerationmodelofemergencyplansforhydraulicengi
neeringprojects.Firstly ,basedonthetextofriskpreventionandcontrolmanualandtheemergencyplan,wepro
poseaknowledgegraphontologymodelofemergencyplan ,constructaknowledgegraphofemergencyplan,and
realisethestructuredexpressionofnon - structuralinformationinthetextofemergencyplan.Secondly,basedon
thewaterresourcesengineeringinspectiontext ,weuseBERT(Bi - directionalEncoderRepresentationfromTrans
formers )+BiLSTM +CRF(Bi - directionalLongShortTermMemorywithConditionalRandomFields)entityrec
ognitionmodeltointelligentlyidentifyriskevents ,projectsandotherentitiesintheinspectiontext.Finally,anin
telligentgenerationtemplateforemergencysolutionsisdesigned ,andthroughmulti - featurefusionofentityalign
menttechnology , knowledge retrievaland inference technology, the intelligentgeneration and pushing of
emergencysolutionsisrealised.Throughthemodelaccuracyanalysisandthevalidationofexamplessuchas“chan
nelleakage ”,thispapershowsahighrecognitionaccuracy(F1valueof96.21%)andareliableemergencyplan
generation,whichcanbeextendedtoemergencymanagementsuchasemergencyrescueofhydraulicengineering
projectsandintelligentgenerationofemergencyplans.
Keywords:knowledgegraph;hydraulicengineering;emergencyplan;South - to - NorthWaterDiversion
(责任编辑:于福亮)
— 6 7 —
6