面向对象的自适应神经模糊-模糊SVM的软件可用性评估(IJIEEB-V5-N1-5)

I.J.InformationEngineeringandElectronicBusiness,2013,1,40-49PublishedOnlineMay2013inMECS()Copyright©2013MECSI.J.InformationEngineeringandElectronicBusiness,2013,1,40-49ObjectOrientedSoftwareUsabilityEstimatewithAdaptiveNeuroFuzzy,FuzzySvmMohammadSaberIraji,ReyhanemosaddeghDepartmentofComputerEngineeringandInformationTechnology,PayameNoorUniversity,[e-mail],[e-mail]—Inthispaper,wepresentmanyintelligentmodelstoestimatetheusabilityofobjectorientedsoftware.Inourproposedsystem,fuzzysvmhaslesserrorsandsystemworkedmoreaccurateandappropriativethanpriormethods.IndexTerms—Usability,objectoriented,fuzzysvm,softwareI.INTRODUCTIONToday,inordertoestimatethequalityofsoftware,softwareindustryexperienceandtrainingdataisimportant.Usabilityisregardedasanimportantqualityofsuccessfulfactorfordevelopingthesuccessfulinteractivesoftwaresystem.Itisalsoakeyqualityfactorinthedevelopmentofsoftwareapplications[1].TheISO9241-11[2]definesusabilityas―thecontextwhichaproductcanbeusedbyspecifieduserstoachievespecifiedgoalswitheffectiveness,efficiencyandsatisfactioninaspecifiedcontextofuse‖.Subsequently,ISO/IEC9126-1[3]categorizedusabilityapartstatinginternalandexternalsoftwarequality,definingitas―thecapabilityofsoftwareproducttobeunderstood,learned,usedandattractedtheuserunderspecifiedconditions‖.AccordingtoE.ChangandT.SDillon[4],theliteraturehasprovideddifferentapproachesforusabilityevaluationwhichare:a)Empiricaltesting[5]b)Inspection[6]c)Comparativeusabilitymeasures[7]d)Formalcomplexity-basedmeasures[8]e)Metricsforusabilitystandardsincomputing[9]Researchstudiesdeterminedthatmanyfactorsinfluencetheusabilityofthesoftwareandthesefactorscontributeintomakinguserperformance,userpreference,anduserinterface[5,7,9,10].Moutinho,Antoniomoraisin2008presentamethodologyforevaluationtheusabilityofthesoftwareforindustrialautomationusingartificialneuralnetworksandconsiderprofileofthesoftware,userprofiles,efficiency,effectivenessandusersatisfactionasfactorsaffectingreusability.Sanjaykumardubey,otherspresentusabilityevaluationofobjectorientedsoftwaresystemusingfuzzylogic,fuzzyAHP[11].Softwareestimationmodelscombiningalgorithmicmodelswithmachinelearningapproaches,suchasneuralnetworksandfuzzylogic,havebeenviewedwithskepticismbythemajorityofsoftwaremanagers[12].Briefly,neuralnetworkstechniquesarebasedontheprincipleoflearningfromhistoricaldata,whereasfuzzylogicisamethodthatusedtomakerationaldecisionsinanenvironmentofuncertaintyandvagueness.However,fuzzylogicalonedoesnotenablelearningfromthehistoricaldatabaseofthesoftwareprojects.Oncetheconceptoffuzzylogicisincorporatedintotheneuralnetworks,theresultisaNeuro-fuzzysystemthatcombinestheadvantagesofbothtechniques[13].However,ourproposedNeuro-fuzzymodelgoesevenfurther:itisauniquecombinationofneuralnetworksandfuzzylogic.Specifically,weobtainedasuiteoffuzzysetstorepresenthumanjudgment,andusedaneuralnetworktolearnfromacomprehensivehistoricaldatabaseofsoftwareprojects.ANeuro-FuzzyusecasesizePointsCalibrationmodelthatincorporatesthelearningabilityfromneuralnetworkandtheabilitytocapturehumanknowledgefromfuzzylogicisproposedandfurthervalidatedinthispaper.Thepaperisorganizedinsixsections.AfterintroducingSectionI,SectionIIwhichalsointroducestherelatedworksofusabilityestimation.SectionIIcontinueswithexplanationsofsupportvectormachineinsectionIII,IV.InSectionIV,VisproposedusabilityestimatewithadaptiveNeurofuzzy,fuzzysvm.Itcontinueswithdiscussionsonthearchitectureofhybridlearningandfuzzymodelvalidation,theerrorofobservationsfortrainingdatasets.SectionVIpresentstheconclusionsoftheresearch.Thepaperendswithalistofreferences.II．SOFTWAREUSABILITYESTIMATEALGORITHMSIn[1]proposedmultiplesub-factorsforthedifferentfactorsofusability.Themultiplefactorsandsub-factorsaredefinedasfollows:i)Effectiveness(A1):Itreferstothecapabilityofthesoftwarewhichusersachievespecifiedgoals.Itcontainsthefollowingsub-factors:a)Accuracy:Itevaluateswhetherthesystem,afterimplementingaspects,isgivingaccurateresultswhenusedunderspecifiedcondition.ObjectOrientedSoftwareUsabilityEstimatewithAdaptiveNeuroFuzzy,FuzzySvm41Copyright©2013MECSI.J.InformationEngineeringandElectronicBusiness,2013,1,40-49b)Speed:Itevaluateshowquicklyataskisperformed.c)Consistency:Itallowsausertoeasilygeneralizehisunderstandingofdifferentmodulesofasystem.d)Understandability:Itdescribesthecapabilityofsoftwaretoenableuserstounderstandtheappropriatenessofsoftwareanditsuseforparticulartasksandconditionsofuse.e)QualityofOutcome:Itevaluatesthequalityoftheinteractionunderstandingorlearningofinformationintheinterface.ii)Efficiency(A2):Itreferstothecharacteristicsoftheproductthatgivesbestresultswithuseofminimumresources.Itcontainsthefollowingsub-factors:a)Scalability:Itistheabilityofasystem,network,orprocess,tohandlegrowingamountofworkinacapablemanneroritsabilitytobeenlargedtoaccommodatethatgrowth.b)Operability:Itdescribesthecapabilityofsoftwaretoenableuserstooperateandcontrolit.c)Compatibility:Itindicatesthataproductcanworkwitho