資源描述:
《] Computational models of trust and reputation Agents, evolutionary games, and social networks》由會員上傳分享,免費在線閱讀,更多相關(guān)內(nèi)容在學(xué)術(shù)論文-天天文庫。
1、ComputationalModelsofTrustandReputation:Agents,EvolutionaryGames,andSocialNetworksbyLikMuiB.S.,M.Eng.,ElectricalEngineeringandComputerScienceMassachusettsInstituteofTechnology,1995M.Phil.,ManagementStudies,DepartmentofSocialStudiesUniversityofOxford,1997Submittedto
2、theDepartmentofElectricalEngineeringandComputerScienceinPartialFulfillmentoftheRequirementsfortheDegreeofDoctorofPhilosophyinElectricalEngineeringandComputerScienceattheMASSACHUSETTSINSTITUTEOFTECHNOLOGYDecember20,2002Copyright2002MassachusettsInstituteofTechnology
3、.Allrightsreserved.SignatureofAuthor:____________________________________________________________DepartmentofElectricalEngineeringandComputerScienceDecember20,2002Certifiedby:___________________________________________________________________PeterSzolovitsProfessor
4、ofElectricalEngineeringandComputerScienceThesisSupervisorAcceptedby:__________________________________________________________________ArthurC.SmithProfessorofElectricalEngineeringandComputerScienceChairman,DepartmentCommitteeonGraduateStudentsComputationalModelsofT
5、rustandReputation:Agents,EvolutionaryGames,andSocialNetworksbyLikMuiSubmittedtotheDepartmentofElectricalEngineeringandComputerScienceonDecember20,2002inpartialfulfillmentoftherequirementsfortheDegreeofDoctorofPhilosophyinElectricalEngineeringandComputerScienceAbstr
6、actManyrecentstudiesoftrustandreputationaremadeinthecontextofcommercialreputationorratingsystemsforonlinecommunities.Mostofthesesystemshavebeenconstructedwithoutaformalratingmodelormuchregardforoursociologicalunderstandingoftheseconcepts.Wefirstprovideacriticalover
7、viewofthestateofresearchontrustandreputation.Wethenproposeaformalquantitativemodelfortheratingprocess.Basedonthismodel,weformulatetwopersonalizedratingschemesanddemonstratetheireffectivenessatinferringtrustexperimentallyusingasimulateddatasetandarealworldmovie-rati
8、ngdataset.Ourexperimentsshowthatthepopularglobalratingschemewidelyusedincommercialelectroniccommunitiesisinferiortoourpersonalizedratingschemeswh