版权说明:本文档由用户提供并上传,收益归属内容提供方,若内容存在侵权,请进行举报或认领
文档简介
InterdisciplinaryContestInofBeItKnownThatTheTeamSijiaJiangYukeZhuRuijieHeWithFacultyShuishengXidianD.ChrisArney,Contest
WasDesignatedOutstandingTinaR.Hartley,TinaR.Hartley,Head Forofficeuseonly
TeamControlC
ForofficeuseonlyMethodsofMeasuringInfluenceUsingNetworkInthispaper,webuildthenetworkmodeltomeasuretheimpactofresearchers,papersandsoon.Wefirstusethenetworkmodeltoevaluatetheimpctofresearchers,consideringresearchersasthenodes,andusingsidestodescribecollaboratonamongresearchers.Wethenproposetheconceptofimportancedegreeandinfluencedegree.Theyareallthepropertiesofnodes.Importancedegreedepictsanode’sownweightinthenetwork,whileinfluencedegreeestimatesanode’stotalnfluencethatmutualimpactamongnodesisincluded.Basedonthecomprehensiveconsiderationoftheclusteringcoefficientanddegree,weputforwardanewideatomeasureanode’simportancedegree.ThencombiningwithPageRankalgorithm,wecanevaluatetheinfluencedegreeofeverynodeinthenetwork.WecanfindALON,NOGAMisthemostinfluentialresearcher.Byyzingtheproprtiesofnodes,wefindthatforageneralresearcher,she/heshouldextenditscollaborativenetworkasgreatlyaspossible,especiallypartnerresearcherswithhighimportancedegree.Second,imitatingpreviousmethodology,webuildthemodeltoevaluatetheimpactofpapers.Weconcludethatfactorsdeterminingapaper’sinfluencecontainthreeindexes:thefirstauthor’sH-value,thejournal’sImpactFactorandcitedindexwhichistheconceptwedefinetodepicttheinfluencedegreeofapaperintheaspectofcitation.Referringtothepreviousidea,weconstructanewnetworkreflectingcitationrelationshipamongpapers,thenwecanobtainthecitedindexofeverypaper.Afterhavingcollectedthevalueofanothertwoindexes,weuseAHPtodeterminetheweightofthreefactors,andfinallyevaluatethetotalinfluenceofapapersuccessfully.WefindthatthepaperStatisticalmechanicsofcomplexnetworksismostinfluential.Afterthatweviewtheotherfieldsinsteadoftheacademicareatoextendourmodel.Weapplyournetworkmodelmeasuremoviestars’influence.WeselectagreatlyinfluentialmoviestartorecethestatusofErdös,andconstructthecooperatingnetworkamongmoviestars.Havingobtainedinfluencedegreeofstars,therankingofinfluenceofmoviestarsmaintainedahighlyconsistentwithreality.Thisisproofthatourmodelisfeasible.Thenwediscussourmodel’sapplicationforacademic,militaryandSNSfieldsroughly.Finally,wemakeasensitivity ysisforourmodel,anddiscusstheimpactofthechangingofnodesandthepapers’feedbackofcitationrelationshipontheresults.Throughprevious ysis,wecanseethatourmodelcanbeappliedtomanyfiles,soithasarelativelyhighgeneralization.. Introduction................................. Background............................... Ourwork................................ Assumptions................................. SymbolDescription............................. Theinfluenceofresearchers......................... Modelone:themeasurementofnodes’importancedegreemodelonnetworktheory................ . ...... EvaluatingtheimportanceofresearcheronlybyNodedegree Evaluatingtheimportancedegreeofresearchersusingclusteringcoefficient............ .............. Theevaluationmodelbasedontheclusteringcoefficientandgree.............. ........... Modeltwo:themeasurementofnodes’influencedegreemodelonPageRankalgorithm. . ................... AnintroductiontoPageRank ................ ThePageRankevaluationmodelbasedontheimportance ysis:someinterestingdata................. Theinfluenceof Thecitingnetworkof prehensveevaluationmodelbasedon Model Applyingmodelstoaspecific ysisofmodels ysisofthecollaborative ysisoftheciting ysisofthe 感谢作者 28747XidianUniversityICM2014OutstandingPaperAdvisior:ShuishengJiangSijia,ZhuYuke,He Copyright2014@All AllRightsPage1ofDoyoubelievethateveryoneintheworldcanestablishcontactwithanyoneelsebyonlysix satmost?EventheU.S.andtheboatmanofVenice,oncefind-ingtheright ,theycanestablishconnection.Thefamoussixdegreesofseparationtheorylsusthataslongaswefindrightmedia,wecanbuildrelationshipbetweenanytwoseeminglyunrelatedentities.Thistheoryisstillapplicableinacademia.Someresearcherscanoftencompletesomehigh-qualitypapersthroughcooperation,whichisinseparablefromthestrongre-lationsamongthem.Inthisera,interdisciplinarystudyisveryprevalent,sotheresearchcapabilityofresearchersandacademicstandardsofthepaperareaffectedbymanyas-pectsofintricateacademicnetwork.Moreandmoreattentionhasbeenpaidtohowtojudgetheacademiclevelandthequalityofpapers.PaulErd¨sisalegendarymathematician.Inhishalf-centurycareerinscientificresearch,hehadmorethan500collaboratorsandpublishedmorethan1400academicpapers.Thereisnodoubtthatheisoneofthemosinfluntialfoundersinthestudyofinterdisciplinary.Peoplecanevendefineaconceptcalled”collaborativedistance”,andErd¨Erd¨Erd¨s2representstheresearcherswhohaddirectcooperationwithErd¨s1,andsoon.TostudytheinfluenceofresearcherswhohaddirectcooperationwithErd¨s(theswhoseErdo¨snumberisone),thispassageisinspiredbyPageRankusedinsearchengineandclusteringcoefficientwhichmeasurestheimportancedegreeofeachnodeinthenetwork,andestablishnetworkmodelbasedongraphtheory.Wegiveamethodologytoevaluatetheinfluenceofresearchersandpapers,andextendthismodeltoanotheraspects.OurFirst,theinfluenceofresearchersandtheimportanceofresearchersarebotharela-tivelyvagueconcept.Inordertogetaclearpictureoftheproblem,wethinkthatthreefactorsshouldbetakenintoconsiderationtomeasuretheinfluenceofoneresearcher:Theresearcher’sextensivedegreeinthefieldofcooperation,namelythenumberofpartners.Thetimesofcooperationwithotherresearcherswhohavestronginfluence.Inthispaper,itisthetimesofcooperationwithErdo¨s.
28747XidianUniversityICM2014OutstandingPaperAdvisior:ShuishengJiangSijia,ZhuYuke,HeRuijie
Page2ofTheacademiclevelofpartnerswhocooperatewiththeThemeasureoftheimportanceofaresearchpapershouldcontainthefollowingTheauthor’sinfluenceoftheresearchpaper’s,wecanmeasureitbyH-Thepopularityofthejournalwhichhavepublishedthisresearchpaper,wecanusethejournal’sImpactFactor(IF)toexpress.ThenumberofcitationbyotherOnthebasisofabovediscussion,toevaluatetheinfluenceofresearchersandresearcherpaper,andtopromoteittobeappliedintheactual,wemayboildownthetaskstothefollowingfourquestions:Bylimitingthesizeofnetworkandextractingthedata,buildacoauthornetworkofthe511researchersfromthefileErdos1andyzethepropertiesofthisnet-thenetwork,andthendoevaluationandrankingforthem.(Erdo¨sisnottheretoytheseroles.)Changetheobjectofstudyanddesignamodeltoevaluatethesignificanceofresearcherpaper.Considerhowyouwouldmeasuretherole,influence,orimpactofaspecificuniversity,department,orajournalinnetworkscience?Dorankingfortheimportanceofresearchpaperandcomparethedifferencebetweenthesetwomethodology.Collectdata,andextendpreviousmodelsandalgorithmstootherfieldsintheactualtoexaminetheiradaptability.DiscussthescienceandutilityofthemodelbuiltIfapaperwascitedmorethanonceinanotherpaper,weregarditasTheimportancedegreewemeasureisforTheaveragevalueofmeasurementindexesofpapersormoviestarscanreflecttheircurrentimpact.ThenumberofcooperationwithErd¨scanaffectaresearchersinfluencedegree,butwhenthenumberexceedacertainvalue,theaffectionwouldbetendtobeaTosomedegree,thequalityofonepaperisproportionaltothenumbercitedbyotherpapers. 28747XidianUniversityICM2014OutstandingPaperAdvisior:ShuishengZhouTeam# JiangSijia,ZhuYuke,HeRuijiePage3ofCopyright2014@AllRightsAllSymbolInthesection,weusesomesymbolsforconstructingthemodelas Theresearcherisimportancedegreeinthe Theresearcherisinfluencedegreeinthe TheH-indexofpaperis TheImpactFactorofpaperisEDeviationDegreebetweentheoldresultsE thenewrankingresultsafterdeletingnodeP.s:OthersymbolsinstructionswillbegivenintheTheinfluenceofBeforemodeling,toavoidambiguitywewilldefinethetwoconfusingTheimportancedegreeofanode:thenode’sinfluenceinthenetwork.Itmea-suresanode’sabilitytocommunicatewithneighbournodesandtobuildthecon-nectionamongdifferentnodesinthewholeTheinfluencedegreeofanode:Duetotheconnectionwithothernodes,thenodewasaffectedbyothernodesandthereforeitpossessesitsowncomprehensiveModelone:themeasurementofnodes’importancedegreemod-elbasedonnetworktheoryInordertomeasuretheinfluencedegreeofeachresearcher,accordingtographtheo-ry,wedefinethenodesasresearchersandthesidesasthepartnershipamong.Thus,weestablishnetworkmodelwhichcanreflecttheinterrelationamongthere-searchers.Tofacilitateunderstandingandfurtherysis,wegivetheschematicdia-gramofthenetworkmodel,showninFigure1.Wecanseethatthroughsmallscopeofcooperation,researcherspromotetheconnec-tivityofthewholenetwork,andcontributetotheconnectionamongseeminglyunrelated Figure1:Thediagramwhichdescribestherelationamong10researchers
Figure2:Afterdeletingnode8,thecon-nectivityofthegraphhasbeendestroyed.Toevaluatetheimportancedegreeofeachresearcherinthenetworkmodel,wediscussitsmeasuringmethodinthefollowingsection.EvaluatingtheimportanceofresearchersonlybyNodeAssumethatnetworkG=(V;E)isundirectedandconsistsof|V|=Nnodes|expressedki=Âj2G<0, :1InFigure1,node8hasthemostnodedegree,ork8=4.Ifdeleteit,asshowninFigure2,node7,9and10 eisolated,theconnectivityofthenetworkarebadlyaffected.Incontrast,itsimpactdegreedeclinelessifthedeletednodesdegreeisless.Therefore,thetyofnodes’degreerepresentstheimportanceofresearchers.But,isthenodeswhichhavethesamenumberofdegreehavethesameWeyzeitinthefollowing,asisshowninFigure3and4Inthepicture,node2and4havethesametiesofdegree.However,deletingnode2hasnoeffectontheconnectivityofthewholenetwork.Afterdeletingnode4,node5isseparated,andnode7,8,9,10losetheconnectionwithnode1,2,3,6.Thusitcanbeseenthatthenodeswhichhavethesamenumberofdegreedonotalwayshavethesameimpactdegree.Sowecanconcludethatnodedegreerepresentsthedirectconnectingabilitywithitsneighbornode,butitcannotreflectitsinfluenceontheconnectivityofthewhole Figure3:Theconditionofconnectionafterdeletingnode2.
Figure4:Theconditionofconnectionafterdeletingnode4network.Tolookforasuitablemethodtosolvethisproblem,weintroducetheconceptofclusteringcoefficient.kAssumethedegreeofnodeiisk,thenthe umtriangleformedbythekneigh-bournodesisC2,hypothesiseeirepresentsthenumberoftriangleformedbyanytwoneighbornodes,thenclusteringcoefficient[4]canbedefinedas:kci=k
Ingraphtheory,aclusteringcoefficientisameasureofthedegreetowhichnodesinagraphtendtoclustertogetherAndinthispassage,itdescribesthecooperationdegreebetweenaresearcherandhispartners.Byyzingthenode2and4intheFigure3and4,wefindbecauseofthenodedegreeofthemareboth3,the umtriangleformedbytheirneighbournodesare33.However,intheactualsituation,thetrianglenumberofnode2is2andnode4is0,soc2=2,c4=0.Theresultindicatesthatothernodeswhichhavecooperatedwithnode4havenoconnectionwitheachother.Node4iscrucialinconnectingothernodesandismoreimportantthannode2.3However,differentfromdegreeindex,clusteringcoefficientcanreflecttheconnec-tivityofneighbournodestosomeextent,butitcannotshowthescaleofneighbournodes.Hence,weshouldevaluatetheimportanceofnodesbyconsideringnodedegreeandclusteringcoefficientsynthetically.TheevaluationmodelbasedontheclusteringcoefficientandFirst,weconsidernodeAssumefiisthesumdegreeofitselfandneighbournodesfornodei,anditcanbeexpressedas:fi=ki+Â Wherekwisthedegreeofnodew,Giisthecollectionofnodei’sneighbournode.reflectstheinformationbetweenthenode’sdegreeanditsneighbourThen,weconsiderclusteringcoefficient.Assumegibe: j=1{fj}-fgi
j}-min{}f}ffwhereciistheclusteringcoefficientofnodei.Becauseclusteringcoefficientindicatetheconnectiondegreeamongneighbournodes,butcannotreflecttheirscale,socicanfinormalized.Shownasequation(4)[5],gialsoshowclosenessamongneighbourAssumepiistheimportancedegreeofnodei,toevaluatetheimportanceofnodessynthetically,wedealwithfipandgibythechemotacticfunction[5]u(x)= ,soweÂgettheimportancedegreepi
si= Nf
Nj=1 j=1Ontheotherhand,considereveryresearcherhasdifferenttimesofdirectcooperationwithErdo¨s,andthetimescanreflecttheimportanceofresearchertosomedegree.Sotofytheindex,weuseapiecewisefunctiontosimulatetheimpactofxdirectcooperationwithErdo¨sontheresearchers’importance,itisexpressedas:8<0.002x11,0<x< 0.2,x�Therefore,thecomprehensiveimportanceofeveryresearcherqi=Pi(x)+ Inordertosimplifythecalculation,weonlystudytheco-authornetworkoftheErd¨s1authors,thenwegettheqiof511researchersandlistthetoptenresearchersasshowninTable1.(Notethatthisrankingisnottheresearcherseventuallyinfluenceranking,buttheimportanceinthenetwork.)Table1:Therankingofresearchers’importancedegreeqCooperation152HARARY,23GRAHAM,RONALDBOLLOBAS,TUZA,FUREDI,SOS,VERAPACH,Bystatistics,Erd¨shadcooperated1671tmesintotalwiththe511researchersandtheaveragenumberofcollaboraivetimeis3yzingthedataaccordingtoTable1,wefindinmostcasesthatthetoptenresearchers’importancearethosewhocooperatedwithErd¨sfrequentlyandtheirfirstcooperationisveryearly.Thereasonableexna-tionforthisphenomenonisthattheirfrequentandearlycooperationhelpthemdevelopandgrowinthecollaborativenetwork,andtheirimportanceisirreceable.Wecallthiskindofgroup”oldresercher”.late(”youngresearcher”)cannotbuildamaturenetworkrelationship.TheirimportanceForexample,ifa”youngresearcher”cancooperatedwitha”oldresearcher”,she/hemaymakeabreakthroughandimprovetheirinfluencedegree.Therefore,itisnotthefinalindextoevaluatetheinfluence.Weneedtoestablishaobjectivemodeltoevaluatetheimpactinviewofthecooperationinfluenceontheresearcher.Modeltwo:themeasurementofnodes’influencedegreemodelbasedonPageRankalgorithmownimportancedegreeandinfluencedegreedeterminedbytheircollaborativenetwork,weintroducePageRankalgorithmtodealwiththisproblem.Becauseofitspowerfulretrievalfunctionandhighqualityretrievalserviceisoneofthemostpopularsearchengines.UsingitsPageRankalgorithmtocalculatethePageRankvaluesofeachwebpage,gettherankofthewebpagebybalancingthenumberoflinkstothesearchtargetandthequalityoftheselinks.ThehigherthevalueofPageRankis,thehighertherankhas[2].ThePageRankevaluationmodelbasedontheimportanceInsection4.1,we’vegotimportantdegreeofeveryresearchers,setaijistheimpor-tancedegreeofnodeitonodej,sowecangettheadjacentmatrixamongnodesinthenetwork,weuseAtoexpress:2 ··· A=6 ···a2n Where,aij
···80,iandj<:qi,iandjdontconnect.(i,j=0,1,2,...,
Todistributetheresearcher’simportanceequallytoresearcherswhocooperated,normalizeeachlineofthematrixsetthisnewmatrixbeA¯,thentransposeA¯andgetthematrixoftransitionprobabilityW,or: AssigneachresearcheraPageRankvaluexiitshouldbedeterminedbyhis(her)partners’importancedegreeqi.Inaotherword,oneresearcher’sinfluencedegreeisproportionaltoher/hispartners’importance.Setthecommonproportionalitycoefficientbel,wecangetthefollowinglinearNÂwijxj=li,(xi=qi,i=1,2,..., LetX=(x1,x2,...,x511)Tbecolumnvectorformedbytheeffectfromotherre-searchers.Throughmatrixmultiplication,equation(4.2.2)canbeexpressedas:WX=l Thuswecanobtainthe umpositiveeigenvaluesofthetransitionprobabilitymatrixxthecorrespondingnon-negativefeaturevectorXmax=(x1,x2,...,x511)T,sowegetresearchers’impactx1,x2,...xnandultimayinfluencerankings.Weonlylisttoptenresearchers,showninTable2.p1ALON,NOGA2GRAHAM,RONALD3BOLLOBAS,04RODL,5FUREDI,06TUZA,07HARARY,8SOS,VERA9SPENCER,JOELFAUDREE,RALPHJASPER, ysis:someinterestingWealreadyknowtheresearchers’importanceindexandinfluenceindexaretwod-ifferentevaluationcrterionsofmeasuringthelevelofscientificresearchers.Theformerreflectstheresearcher’sabilitytocontributetotheconnectioninthenetworkbycontact-ingotherresearchers,whilethelattershowstheresearcherisaffectedbyherself/himselfandher/hispartnersandcanchangethesefactorsintoher/hisoverallinfluence.Comparingtherankresultofthetwomethods,wefindthatthechangeofoverresearchers’rankingiswithin75.Inviewoftheirdifferenceofemphasis,itisHowever,yzingtheremainingdata,wefindsomeinterestingphenomenon,weshowitinTable3Intheabovetable,wedefinetheareaofyellowasstabledata,theareaofgreenaspositivedataandtheareaofredasnegativedata.Stabledata:we yzethetopfiveinfluenceofresearchersandfindtheirrankingiscloselytotheirimportanceranking.WeconcludetheinfluencedegreeoftheresearcherswithlargeimportancedegreeisalsoTable3:Theschematicdiagramofresearchers’
p- 1105GRAHAM,RONALD321BOLLOBAS,532440FUREDI,752PYBER,3TETALI,PRASAD22FISHBURN,PETERTENENBAUM,7STEIN,ALAN1SMITH,PAUL1MAXSEIN,11Positivedata:Someresearchers’influencedegreerankingimprovebymorethan250thantheirimportancedegreeranking,someevenimproveby351.TheircommonfeatureistheirfirstcooperationwithErdo¨sislate,sotheircooperationnetworkcannotdevelopmaturelyandtheirimportancedegreerankingislow.However,byooperaingfrequentlywithhighinfluenceresearchers,theirinflu-encedegreecanbepromoted.TakeFISHBURN,PETERCasanexample,noton-searcherswhoseimportancedegreeishigh(FUREDI,ZOLTAN,GRAHAM,RONALDLEWISandSPENCER,JOELHAROLD),sohisinfluencedegreerankinghavegreatlyimproved.Weconcludethatevenwithlowerimportancedegreeinthenetwork,onecanenhanceitsinfluencedegreebycooperatingwithhighlyin-fluentialresearchers.Negativedata:Wefindfourresearchers’influencedegreedropmorethan100.Investigatingtheirdata,wegettheirpartnersaremorethan20,butoverhalfofthemdon’tbelongtoErdo¨s1.Theresultshowsthattheresearcherwhocooperatelessfrequentlyhavelowinfluencedegree.Sowecandrawtheconclusionthataresearchercanimprovetheirinfluencebyenhanc-ingitsowncollaborativenetwork,andtheimportancedegreeofitspartnersysTeam#
28747XidianUniversityICM2014OutstandingPaperAdvisior:ShuishengJiangSijia,ZhuYuke,He Page11ofCopyright2014@AllRightsAllimportantroleinenhancingitsTheinfluenceof
Toevaluateresearchpapers’influence,wechoose15researchpapersinandusethemethodofthesection§4.Weestablishamodelwhichreflectsthein amongresearcherpapers.Undoubtedly,thebestwaytoevaluatetheinfluencedegreeofaresearchpaperisexaminingthequalityofthecontent.However,theredoesn’texistagreatanddirectmethodforit.Therefore,weneedtolookforindexestoevaluateindirectly.By-ysis,wethinkthefactorsthataffecttheinfluencedegreeofresearchpapersismainlymanifestedintwoaspects.Externalinfluencedegree:thecitingofapaperisanimportantindextomeasureitsinfluence.Byyzingtherelationofthemutualcitingofthese15papers,wecanbuildacitingnetworksimilartomodelone,andusePageRankalgorithmtoevaluateeverypaper’sinfluencedegree.Internalinfluencedegree:Itisalsoaffectedbyitsfirstauthor’slevelofscientificresearch,namelyHindex,andtheinfluenceofthejournal,namelyImpactFactor(IF).Sowechoosetheabovethreeindexestoevaluatepapers’influenceThecitingnetworkofFirstofall,differenfromnetworkrelationshipofresearchers,oneresearchpapercanonlycitethepaperpublishedbeforeitandthereisnorelationshipofcooperationbetweenthem.Nevertheless,oncected,itindicatesthisresearchpapergetstheaffirmationfromotherresearchers.Themoretheresearcherpaperiscited,thehigherinfluenceofthisresearcherpaperwillhave.Meanwhile,thequotercanalsobenefitfromitandimprovehis(her)paperimpact.Hence,wecanestablishfeedbackrelationshiplikeresearchers.Weestablishthenetworkrelationshipgraphamongresearchpapers.Thenodesrepresentresearchpapersandthesidesrepresentthefeedbackrelationshipamongresearchpapers(weassumethattheeffectsofinctionoftwopapersexistcitingrelationareequal),shownintheFigure5.UsingthePageRankalgorithminmodeltwo,wecangettheinfluencedegreeinthenetwork,wecallitpapercitedindex,setitber,whichmeasuresitscitedinfluence.Bycollectingdatafromauthoritativewebsite(suchasSCI,Scholar),weusethenumbercitedbyothersastheindextomeasuretheimportancedegreeqofeveryTeam#
Figure5:Thecitationnetworkamong
Page12ofSettheadjacencymatrixofthenodesinthenetworkbe ··a ···A=4
·<80iandjconnectThereinto,aij=:<q,iandjdon’tNormalizeAanduseequation11),wecangetrofthese15research
prehensiveevaluationmodelbasedonrcanonlyreflecttheeffectfrometernalelement.Toevaluatearesearchpaperover-all,wearedeterminedtouseAHPandcalculatetheweightofIF,Hindexandr.Wethinkthatintheprogressofmeasuringaresearchpaper,thefirstauthorshouldn’tbeconsideredtoomuch.Exceptthat,thecontributionofinternalfactorisgreaterthanexternalfactor.Sowesetthesizeofrelationshipofthethreefactors:IF>r>H.Bythisprinciple,wegivePairwisecomparison0 r B 4 1@1 Team#
Page13ofWetheninputthematrixintoYAAHPandcalculatetheweightofeachfactor:aIF0.5469,aH=0.1085,ar=Thefinalexpressionofevaluatingonepaper’sinfluencedegreeTi=aIFIFi+aHHi+arri Inthefollowing,wetesttheconsistencyoftheAHP.TheconsistencyindexCI=lxnshouldbecloseto0;wegetCI=TheconsistencyratioCR=CIshouldbelessthan0.1;WegetCR=0.02.ourdecisionmethoddisysperfectlyacceptableconsistencyandweightsareListthetopfiveinfluencedegree,showninTableTable4:Theschematicdiagramofpaprs’ r12 24 3 4 -57 -Inthefirstthreetasks,wehavesolvedtheproblemofmeasuringtheinfluencedegreeofresearchersandpapers.Ifweareabletogetenoughinformationaboutresearchpapersandresearchers,wecancreatearesearchdatabase.Ifwewanttoevaluatetheresearchstrengthofauniversityoradepartmentinacertainfield,weneedtolookfortheresearchersandtheresearchpapersinrecentyearsofthisorganization,theninquiretheirinfluenceinthedatabase.Finallywecandeterminethestrengthofitbychoosingasuitableevaluation(suchasTOPSIS).ModelInmodeloneandmodeltwo,usingthenetworkmodel,wefytheinfluenceofresearchersandpapers,showingthestrongabilityofourmodeltoevaluateacademicindex.Asamatterofafact,ourmodelscanbeappliedinvariousfieldsseeminglyhavenothingtodowithscience.Sointhefollowingsection,wewillemployournetworkmodeltoevaluatetheinfluenceof moviestars,andconsideritsextensionwidely. Copyright2014@All All Applyingmodelstoaspecific
Page14ofThecinemaofHongKongisoneofthethreemajorthreadsinthehistoryofChi-neselanguagecinema,alongsidethecinemaof,andthecinemaof.Fordecades,HongKongwasthethirdlargestmotionpictureindustryintheworldandthesecondlargestexporter.SoitisobviousthattheindustryofHongKong’scinemayssuchapredominateroleinandeventhewholeworldthatithasgreatlyprompt-edthedevelopmentofthecinemaindustryinand.TherearesomeilluminatedmoviestarsfromHongKong,suchasJackieChan,Tonyleungchiuwai,AndyLauandsoonwhoalsohaveagreatreputationaroundtheworld.[6]ForthefactthatTonyleungchiuwaihasaverymagnificentcinemacareer,includingbeingawardedtheGoldenHorsebestactornominationmorethanonce,wedecidetorecetheErdo¨sinthepreviousnetworkmodelwithhim,andthenqualifytherangeoftimeandtheregion,inorderingthatwecanconstructanewnetworkmodelbasedontheModel1whichcanbeusedtoassessamoviestar’sinfluencedegree.ForthefactthatTonyleungchiuwaihasaverymagnificentcinemacareer,includingbeingawardedtheGoldenHorsebestactornominationmorethanonce,wedecidetorecetheErdo¨sinthepreviousnetworkmodelwithhim,andthenqualifytherangeoftimeandtheregion,inorderingthatwecanconstructanewnetworkmodelbasedontheModel1whichcanbeusedtoassessamoviestar’sinfluencedegree.Besidesthat,wealsoneedanindextoassesseachactor’seffectonthe’sbox-office,andtheindexcanbeconsideredastheimportancedegreeModel1asabalanceofaactor’sstrength.Wenormalizetheaverageofthebox-officeofmovieswhereeachactorhasprticipatedinrecentfiveyearsasthenumberofthisbox-office’sinfluence.Table5:Theindexesofthemovieinfluencebox-1Zhou29Zhao34Andy5Zhang6Shu78Jet89Donnie98Jackie8LouisTeam
Page15ofPuttingthenumberintothematrixdepictingthecooperatedrelationshipbetweendifferentactors,wecangetthenumericaldataofthe31actorsinfluencedegreeusingthePageRankalgorithm.Rankingthestarsbytheinfluencedegree,wecangettheToptenshowninthechart:yzingtheTable5,wecanfindoutthatthebox-officeindicatingtheactorsstrengthhassomecontributiononthestarsinfluence,buttheactorscooperatingrela-tionshipnetworkalsoisanunignorablefactor.TakingZhangZiyiasanexample,whileshehasainferiorbox-officethanotherstars,wejustfindthecorrespondingnodehasahigherdegree,thatis,shehasmanymorecooperatedstarsthanothers,whichresultsinherhigherinfluencedegree.However,theJackieChanisanoddcase,hemainlypartici-patedintheactionmovies,addinghislackofcooperationwith overallinfluenceisnotthatsuperiordespiteofhisstrongbox-office.What’smore,thecurr
温馨提示
- 1. 本站所有资源如无特殊说明,都需要本地电脑安装OFFICE2007和PDF阅读器。图纸软件为CAD,CAXA,PROE,UG,SolidWorks等.压缩文件请下载最新的WinRAR软件解压。
- 2. 本站的文档不包含任何第三方提供的附件图纸等,如果需要附件,请联系上传者。文件的所有权益归上传用户所有。
- 3. 本站RAR压缩包中若带图纸,网页内容里面会有图纸预览,若没有图纸预览就没有图纸。
- 4. 未经权益所有人同意不得将文件中的内容挪作商业或盈利用途。
- 5. 人人文库网仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对用户上传分享的文档内容本身不做任何修改或编辑,并不能对任何下载内容负责。
- 6. 下载文件中如有侵权或不适当内容,请与我们联系,我们立即纠正。
- 7. 本站不保证下载资源的准确性、安全性和完整性, 同时也不承担用户因使用这些下载资源对自己和他人造成任何形式的伤害或损失。
最新文档
- 果品综合检测财务报表预测
- 农资营销半年工作总结(5篇)
- 土木地质实习报告
- 关于质量安全承诺书范文集合六篇
- 高考数学复习解答题提高第一轮专题复习专题01数列求通项(数列前n项和Sn法、数列前n项积Tn法)(典型题型归类训练)(学生版+解析)
- 专题8.1 统计和概率的简单应用(例题讲解)-2022-2023学年九年级数学下册基础知识专项讲练(苏科版)
- xx学校加强预防未成年人被侵害和违法犯罪工作的实施方案
- 语文统编版(2024)一年级上册识字5 对韵歌(新) 教案
- 华师《现代汉语语法与修辞》在线作业
- 部编版历史八年级上册第八单元 第26课《教育文化事业的发展》检测卷(后附答案及解析)
- 数学初一上学期数学期末模拟试卷带答案
- 二年级上册数学家长会 课件 (共16张PPT)
- DB11-T1727-2020火灾后钢结构损伤评估技术规程
- 多彩的活动 作文批改评语
- 新人教版小学数学六年级上第六单元备课方案教学设计教案
- 爱惜学习用品 完整版课件
- 广东省普通高等学校毕业生就业推荐表
- 啤酒销售表格明细模板
- 黄鹤楼初中古诗词教案PPT课件讲义
- 电大中级财务会计二形考任务1-4答案
- 2022年数据结构大作业题目
评论
0/150
提交评论