基于复杂网络的ICT产业新产品扩散模型研究

上海交通大学硕士学位论文基于复杂网络的ICT产业新产品扩散模型研究姓名：郭浩申请学位级别：硕士专业：管理科学与工程指导教师：陈忠20090101IVICT20%ICTICTICTICTICTICTICTSWSFICTICTVNEWPRODUCTDIFFUSIONMODELOFICTINDUSTRYBASEDONTHECOMPLEXNETWORKSABSTRACTICTindustrytakesmorethan20%shareofChina'sGDP,anditplaysavitalroleinthenationaleconomicdevelopment,forICTindustryhasbecomeanimportantplatformforthecountry'scompetitivenessupgrading.ThenewproductofICTindustryhasaverystrongnetworkeffectscharacteristicandthusthereliesagreatdifferencebetweenthediffusionmodelofnewproductofICTindustryandthetraditionaldiffusionmodelofnewproduct.ThefocusofthisdissertationisthediffusionmodelingproblemofnewproductsofICTindustry.Thisstudyofthisdissertationincludesthenetworkeffectaccordingtothecharacteristicsofthisindustry,theanalysisoflocalandglobalnetworkeffects,andtheimportantroleofnetworkeffectswhenconsumersmakingtheirdecision.Anagent-basedsimulationmodelisestablishedtoanalyzetheinfluencebynetworkeffectsandstructuretothediffusionofnewICTproductsandcomparethediffusionpatternsunderdifferentnetworkeffects.Weputforwardthatwheninstalledbaseistoosmall,highernetworkeffectsintensitymayleadtoworsediffusionpattern,andglobalnetworkeffectcaneasethisVIinfluence.AlsoweputforwardthatSWnetworkcanbenefitconsumersincommunicationandraisediffusionextentaccordingly,andthescale-freecharacteristicsofSFnetworkcanbeusedtomarketingactivities.AlltheseconclusionsnotonlyindicatethenetworkeffectcomposedofconsumerinteractionsisanimportantfactorinfluencingthediffusionofICTnewproducts,butalsoprovideintervenestrategiesforenterprises.KEYWORDS:diffusiontheory,networkeffect,complexnetworks,ICTindustryIIIII1(InformationandCommunicationTechnologyIndustryICT)(NetworkMarket)Frels[1]3GCDDVDFirefoxGoogleInternetExplorerLive2ICT20%ICT2000ICT10%200427%ICT19962004ICT50%3750ICT35032901000%2004ICT46%1800ICT14902005ICT30%ICTICT20053.03——8100——270ICT20051180ICT6ICT20061.37200640%200540200613701040“”ICTICTICTChakravorti[2]3ICTICT12ICTSICTSSICTICT214Authur[3]206020702080SantaFe205ICT+?ICT6ICTICTICTICTSWSFICTICTICT78ICT1912(innovation)Rogers[4][5]Windows9Robertson[6](DiscontinuousInnovation)(ContinuousInnovation)(DynamicallyContinuousInnovation)[7](word-of-mouth)[8],Rogers[4]“”Robertson[6]12310[9]Rogers[10]12345[11]SS[8]11SSSS[8]St——(Takeoff)(Slowdown)——SSFigure:TheS-StyleDiffusionCurve10%25%34%50%-96%15%-32%GNP12[4]Bass[12][13]Bass[4](CellularAutomata)(Agent-basedModels)(PercolationModels)13(DiffusionRate)(),dyfytdt=()yyt=fyt()()()()[]dNtntgtMNtdt==-………….………………(2-1)()ntt()NttM()[]MNt-t()NtM()gtt()gt()gt()Nt()()gtabNt=+………………………………..(2-2)()(())[-()]ntabNtMNt=+………………………….(2-3)aba12b[-()]aMNt()[-()]bNtMNt14(Agent)(b=0)()()[-()]dNtntaMNtdt==………….……………(2-4)(a=0)()()()[-()]dNtntbNtMNtdt==…………………….(2-5)(0,0ab≠≠)()()[()][-()]dNtntabNtMNtdt==+…………………..(2-6)[14]FourtWoodlockShusterIPMansfield(Logistic)15BemmaorGompertzRouvinenBassBassBassg(t)BassBassBasst()()()1-ftpqFtFt=+……………………………(2-7)()Ftt()ft()Ftttpq0p=Bass0q=Basst()ftt()FtpqM()ntt()()ntMft=()Ntt16()()NtMFt=Bass()()()NtntpqMNtM⎛⎞=+-⎡⎤⎜⎟⎣⎦⎝⎠………….………….(2-8)()gt()NtpqM+BassBass1tS2pqpq3Bass0p=[16]0q=[15]BassBass[17]KamakuraBalasubramanianHorskySimontJonesRitz—BassJainMahajanMullerBassMahajanSharmaBuzzellKrishnanBassKumarParkerGatignon[18]BucklinSenguptaGupta,JainSawhney17BassBassBass(IndividualInteraction)(DiffusionExtent)Granovetter[19]Bikhchandani[20]Banerjee[21](InformationalCascades)(HerdBehavior)Morris[22]18Jensen[23][24]ConleyUdry[25]Young[26]“”Kapur[27]GaleKariv[28]EllisonFudenberg[29][30]BalaGoyal[31]BalaGoyal[32]“”(BandwagonEffect)(SnobEffect)(WordofMouth)Leibenstein[33]“”AbrahamsonRosenkopf[34][35]Thun[36]BassAlkemadeCastaldi[37]19,[38][39]EllisonFudenberg[30]20FloryRapoportErdosRényi[40][41][5]2µ——(21)2Figure2:Themappingsfromacomplexnetworktoafeaturevector.(Symmetry)(Thresholding)3Figure3:Thefourmaintypesofcomplexnetworksandtheirtransformations.GNN(G)nE(G)1,2,...iN=(,,)ijwijw123Mmmmmm⎡⎤⎢⎥⎢⎥⎢⎥=⎢⎥⎢⎥⎢⎥⎣⎦rM22(,,)ijw11(,,)ijw,22(,,)ijw12ii≠12jj≠“”(,10)ij(,)ij()(,)ijijjiWijw(,)ijWA()TAWd=ijwT1ija=,0ija=A()T=+(TWW)0ija≠ij(Predecessor)(Successor)0ija≠jiiji()iniiki()in(“”)outikinikoutiniiikkk=+iiSiSioutiijjiniijiSwSw==∑∑……………………………….(3-1)23(Degree)(DegreeDistribution)(Connectivity)P(k)kP(k)k/()~kPkeb-b()~Pkkg-g[5]2-11Table1:Thedegreedistributionindexofseveralrealnetworksg449913255164822.347000000800000002.1Email59912863001.5/2.0Nokia.com501585000002.2/2.424578??2.151248323473261831832.01/2.8347000000800000002.1[5]ijm()()()1121,,,...,,mikkkkj-ij(Path)m(Distance)ijijijijijd25ijd=∞ijjidd≠L()11ijijLdNN≠=-∑……………………………(3-2)(GlobalEfficiency)()111ijijENNd≠=-∑……………..…………….(3-3)LatoraMarchiori(ClusteringCoefficient)[40]kikk()12kk-iEi:()21iiECkk=-………………………………..(3-4)261iiCCN=∑…………………………………(3-5)1C=k()1iikCkCn=∑………………….……………(3-6)knkkICTICTErdosRényiRapoport[5]ERNPNP()1kpN=-P(N→∞)kkPN1kPN=-K()()111NkkNPkppk---⎛⎞=-⎜⎟⎝⎠…