自动化构建的中文知识图谱系统2

2015-09-062015-11-12。9732014CB34040414511108002。1991—CCF、、1993—、1962—CCF、、、。1001-9081201604-0992-05DOI10．11772/j．issn．1001-9081．2016．04．0992*201804*[e-mail]、。。CayleyMongoDB。YAGO、HowNet50%。TP311．5AAutomaticalconstructionofChineseknowledgegraphsystemEShijia*LINPeiyuXIANGYangCollegeofElectronicsandInformationEngineeringTongjiUniversityShanghai201804ChinaAbstractTosolvetheproblemthatthemethodscurrentlyusedtoconstructChineseknowledgegraphsystemaretime-consuminghavelowaccuracyandrequirealotofmanualinterventionanintegratedend-to-endautomaticallyconstructedsolutionbasedonrichdatafromChineseencyclopediawasproposedandauser-orientedChineseknowledgegraphwasimplemented．InthissolutionsomepropertyandrelatedtextinformationoftheoriginalencyclopediadatawerescrapedtolocalsystemuninterruptedlybythecustomWebcrawlerandsavedasatriplewithextendedattributes．Throughgraph-orienteddatabaseCayleyanddocument-orienteddatabaseMongoDBthedatainthearchivedtriplefileswasimportedintheback-endsystemandthenconvertedtoahugeknowledgegraphsysteminordertoprovidevariousservicesdependentontheChineseknowledgegraphinthefront-endsystem．Comparedwithotherknowledgegraphsystemstheproposedsystemsignificantlyreducestheconstructiontimemoreoverthenumberofentitiesandrelationsisatleast50%higherthanthatoftheotherknowledgegraphsystemssuchasYAGOHowNetandtheChineseConceptDictionary．KeywordsknowledgegraphWebcrawlertriplefileknowledgebasegraph-orienteddatabase0、12。InternetMovieDatabase、YAGO3－4、DBpedia5－6Freebase7。8。9－10。、11－1213。。。。、、。JournalofComputerApplications2016364992－9961001ISSN1001-9081CODENJYIIDU2016-04-10http//．joca．cn1C1C2…CnCiIiＲ1Ｒ2…Ｒm。“is-a”AB“”“”。“is-a”Ci。“is-a”A→BBA。1。1。AA。。。2。。2．12．1．1PythonScrapy14、2。2UＲL。HTMLUＲLUＲL。UＲL。。2．1．23。3、。、。XXX。1。Tarjan15。。“”“”。“”“”。4。。42．2“”“”。16。“”“”。17。。3994。。。。。。“”“”———“”。———〈〉“”。〈12〉。2．2．1InfoBoxInfobox3。Infobox。5InfoBox“”InfoBox。5InfoBox1〈〉2〈computer〉3〈1946〉4〈ENIAC〉5〈〉6〈··〉。2．2．2HTML6。“”“”《》。〈〉。《》“”“”《》。HTML、InfoBox。62．3。。。。。Infobox〈〉、、。。。。。。33．1〈SubjectPredicateObjectLabel〉。Label。Label。LabelLabel。3．2。7。。CayleyMongoDB。MongoDBCayleyCＲUDCayleyMongoDB。MongoDBCayley“quads”。2。Predicate1Predicate“is-a”“instance-of”499362Predicate“attribute”SubjectPredicateObject3Predicate。7InfoBox。MongoDB。。74。。WebContentUnderstanding-KnowledgeGraphCU-KG。4．1CU-KG。8。8b“”“”。“”。4．2CU-KG、。Predicateattribute。〈attributeUＲL〉“”UniformＲesourceLocatorUＲL。、9。894．3。CU-KG“”“2012-08-10”“2012-08-13”“2012-08-10”10。10“”“2012-08-10”1d115994“2012-08-13”。11“”“2012-08-11”。4．4。、。“”〈〉、〈〉、〈〉“”。。5。。。。。Freebase、WordNet。。CU-KG。、。。。CU-KG1。1CU-KGWordNet155287479887YAGO10546385000000HowNet19192446243366590—CU-KG339961556516141CU-KG50%。CU-KG。12．5h18。CU-KG、。6CU-KG。。CU-KGNoSQL。CU-KG。1LENATDB．CYCAlarge-scaleinvestmentinknowledgeinfra-structureJ．CommunicationsoftheACM1995381133－38．2SINGHALA．IntroducingtheknowledgegraphthingsnotstringsEB/OL．2014-10-10．https//googleblog．blogspot．com/2012/05/introducing-knowledge-graph-things-not．html#/2012/05/introducing-knowledge-graph-things-not．html．3SUCHANEKFMKASNECIGWEIKUMG．Yagoacoreofse-manticknowledgeC//Proceedingsofthe16thInternationalCon-ferenceonWorldWideWeb．NewYorkACM2007697－706．4SUCHANEKFMKASNECIGWEIKUMG．Yagoalargeontol-ogyfromWikipediaandWordNetJ．WebSemanticsScienceServicesandAgentsontheWorldWideWeb200863203－217．5AUEＲSBIZEＲCKOBILAＲOVGetal．DBpediaaNucleusforaWebofOpenDataM．BerlinSpringer2007722－735．6BIZEＲCLEHMANNJKOBILAＲOVGetal．DBpedia—acrystallizationpointfortheWebofdataJ．WebSemanticsSci-enceServicesandAgentsontheWorldWideWeb200973154－165．100169936MapＲeduce、。PrePost。1AGＲAWALＲIMIEILNSKITSWAMIA．MiningassociationrulesbetweensetsofitemsinlargedatabasesC//Proceedingsof1993ACMSIGMODConferenceonManagementData．NewYorkACM1993207－216．2AGＲAWALＲSＲIKANTＲ．FastalgorithmsforminingassociationrulesC//VLDB1994Proceedingsofthe20thInternationalConferenceonVeryLargeDataBases．SanFranciscoMorganKaufmannPublishers1994487－499．3LINKCLIAOIECHENZS．AnimprovedfrequentpatterngrowthmethodforminingassociationrulesJ．ExpertSystemswithApplications20113855154－5161．4GUPTAＲSATSANGICS．AnefficientrangepartitioningmethodforfindingfrequentpatternsfromhugedatabaseJ．InternationalJournalofAdvancedComputerＲesearch20122262－69．5．FP-treeJ．2011311101－103．LIYBTANGHHEYM．FrequentpatternminingalgorithmbasedonimprovedFP-treeJ．JournalofComputerApplications2011311101－103．6SUCAHYOYGGOPALANＲP．CT-PＲOabottom-upnonre-cursivefrequentitemsetminingalgorithmusingcompressedFP-treedatastructureC//FIMI2004ProceedingsoftheIEEEICDMWorkshoponFrequentItemsetMiningImplementations．Piscat-awayNJIEEE2004212－223．7ZAKIMJGOUDAK．FastverticalminingusingdiffsetsC//Proceedingsofthe9thACMSIGKDDInternationalConferenceonKnowledgeDiscoveryandDatamining．NewYorkACM2003326－335．8LIZFLIUXFCAOX．AstudyonimprovedEclatdataminingalgorithmJ．AdvancedMaterialsＲesearch2011328/329/3301896－1899．9DENGZHWANGZHJIANGJJ．Anewalgorithmforfastmin-ingfrequentitemsetsusingN-listsJ．ScienceChinaInformationSciences20125592008－2030．10LINKCLIAOIECHANGTP．Afrequentitemsetminingal-gorithmbasedontheprincipleofinclusion-exclusionandtransac-ti