大数据群体计算中用户主题感知的任务分配.pdfVIP

计算机研 究 与发展 DOI：10．7544／issn1000—1239．2015 JournalofComputerResearchandDevelopment 一 52(2)：309—317，2015 大数据群体计算中用户主题感知的任务分配 张晓航 。李国良。冯建华 (清华大学交叉信息研究院 北京 100084) 。(清华大学计算机系 北京 100084) (zhangxiaohang12@mails．tsinghua．edu．cn) Theme-AwareTask AssignmentinCrowdComputingonBigData ZhangXiaohang ，LiGuoliang，andFengJianhua。 (InstituteforInterdisciplinaryInformationSciences，TsinghuaUniversity，Beijing100084) 。(DepartmentofComputerScienceandTechnology，TsinghuaUniversity，Beijing100084) Abstract Bigdatahasbroughttremendouschallengesforthetraditionalcomputingmodel，becauseof itsinherentcharacteristicssuch aslargevolume，high velocity，high variety，low—density value．On theonehand，thelargevolumeandhigh velocityrequirethetechniquesofmassivedatacomputation andanalysis；ontheotherhand，thehighvarietyandlow—densityvaluemakebigdatacomputingtasks highlydependonthecomplexcognitivereasoningtechnology．Toovercomethecoexistencechallenges ofmassivedataanalysisandcomplexcognitivereasoning，human—machinecollaborationbasedcrowd computingisaneffectivewaytosolvethebigdataproblem．Incrowdcomputing，taskassignmentis oneofthebasicproblems．H6weVerthecurrentcrowdsourcingplatform canndtsupporttheactive task assignment，which iteratively assignstasksto appropriateworkersbased on theknowledge backgroundorusers．Toaddressthisproblem ，weproposeaniterativetheme—awaretaskassignment framework，and deploy it into existing crowdsourcing platforms．The framework includes tWO components．Thefirstcomponentistaskmodeling，whichmodelsthetasksasagraphwherevertices aretasksandedgesaretask relationships．Thesecond componentistheiterativetaskassignment algorithm ，which identifiesthe themesofthe workersby their historicalrecords，computesthe workersaccuracy on differentthemes，andassignsthe tasksto theappropriateworkers．Various experimentsvalidatetheeffectivenessofourmethod． Keywords cr