摘要
Forworkflow-basedservicecompositionapproach,therelationsbetweentheWebserviceQoSandenvironmentsareusuallynotconsidered,sothattheinformationaboutQoSforcompositeserviceselectionisinaccurate.Itmakestheselectedcompositeserviceinefficient,orevenunexecutable.Toaddressthisproblem,anovelservicecompositionapproachbasedonproductionQoSrulesisproposedinthispaper.Generally,itisverydifficulttodirectlyanalyzehowdifferentkindsofenvironmentfactorsinfluencetheWebserviceQoS.Weadopt'black-box'analysismethodofoptimizingcompositeservices,discoveringtheknowledgesuchas'theQoSofoneWebservicewillbehigherinspecificenvironments'.Inourapproach,theexecutioninformationofthecompositeserviceisrecordedintoalogfirst,whichwillbetakenasthebasisofthesubsequentstatisticalanalysisanddatamining.Then,thetimelyQoSvaluesoftheWebservicesareestimatedandtheproductionQoSrulesbeingusedtoqualitativelyexpressthedifferentperformancesoftheWebserviceQoSindifferentenvironmentsaremined.Atlast,weemploytheminedQoSknowledgeoftheWebservicestooptimizethecompositeserviceselection.Extensiveexperimentalresultsshowthatourapproachcanimprovetheperformanceofselectedcompositeservicesonthepremiseofassuringtheselectingcomputationcost.
出版日期
2010年06月16日（中国期刊网平台首次上网日期，不代表论文的发表时间）
