外文翻译---问题解决和决策.doc
Selection1.1PROBLEM-SOLVINGANDDECISION-MAKEINGItishelpfulatthisstagetointroducetheideasofproblemsolving,astructureforwhichwasfirstdescribedbyDewey(1910).ThestagesidentifiedbyJohnDeweywere:Whatistheproblem?Whatarethealternatives?Whichalternativeisbest?YoushouldnowbeinapositiontoidentifyaresemblancebetweenDeweysthreestagesandthesoftwarelifecycle.Enoughhasbeensaidaboutthelifecycleearlierforustospotadistinctresemblancebetweenthefirststageofproblemdefinitionandourrequirementsanalysisphase.Infactmanyorganizationsusethetermproblemorprojectdefinitionratherthanrequirementsanalysis.Thefinaltwostagesmaysimilarlybeidentifiedasbeingequivalenttothephasethatwerefertoasdesign.Amorerecent,butveryrelevantstructurehasbeensuppliedinthecontextofdecision-making(Simon,1960).ProfessorSimonlabelsthestagesofdecision-makingasfollows:Intelligenceactivity,DesignactivityandChoiceactivity.ThetermIntelligenceisusedhereinthemilitarysense,i.e.,searchingtheenvironmentforconditionsrequiringadecisiontobemade.Designisconcernedwithinventinganddevelopingpossiblecoursesofaction.Theactivityofselectingaparticularcourseofactionfromthoseavailableisknownaschose.Inthiscaseourrequirementsanalysiscorrespondstotheintelligenceactivity.Althoughthesoftwaredesignerdoesnotneedtosearchtoohardforcircumstancesrequiringadecisiontobemade,theywillnormallybefoundlyingonhisorherdesklabeledSpecificationofrequirement.HoweverSimonsuseofthetermDesigndiffersfromours.WhereasSimonusesthewordtodescribethegenerationofpossiblesolution,weuseitinthesensethatselectionisincludedalso.Thereseemstobesomejustificationforbelievingthatproblem-solving,decision-makingandsoftwareanalysisanddesignshareacommonframework.Anditisnotstretchingthepointtoofartoclaimthatthefirsttwoare,ineffect,exactlythesamethingwhilethelastisjustaparticularinstanceofthisphenomenon.Consequently,weshallpersistinregardingsoftwaredesignasaproblem-solvingactivityandtreatingitassuch.Thismeansthatwemustdevotesomeattentiontothequestionsofgeneratingpossiblesolutions(i.e.,designs),andtheselectionofonefromthisnumber.1.2SIZEOFSELECTIONLetusstartwithaverysimpledesignproblem.AsoneoftheparentsofanuclearfamilyyoudecidetakeyourspouseandyourtwoandahalfchildrentoScarboroughfortheday.Yourdesignproblemistoestablishthebestwayofmakingthejourney.Youestablishthealternativesasbeing:BritishRail,NationalBusorprivatecar.Inordertomakeyourchoiceyouneedsomethingelse.Youcannotpossiblydecidewhichisthebestunlessthereissomepropertypossessedbythethreealternativeswhichisimportanttoyouandwhichyouwishtoseemaximizedorminimized.Hence,ifyouwishtominimizethecostoftheouting,say,thedecisionissoonmade,givensomeunderstandingofthefarestructuresofBritishRailandNational,andarealisticappraisalofthefuelconsumptionofyourcar.Suchaproperty,inthiscasethedirectcostofthejourney,isknownasthedesigncriterionordesignobjective.Similarly,youmighttakejourneytimeasthedesigncriterionandonceagainthestudyoftimetablesandknowledgeofyourcarscapabilitieswouldenableareadychoicetobemade.Inpassing,weshouldnotethatifbothcostsandjourneytimeareimportantthenthechoicemightbemoredifficult.Butthelatterpointisforfuturedevelopment.Forthemomentwemustconcentrateonsizeofselection.1.2.1CombinatorialexplosionIntheforegoingexamplethedesignproblemwastrivial,fortheselectionhadtobemadefromthreealternatives,eachoneofwhichwaseasilyevaluated.Butnowrecallthatsometimeinthepastweaskedyoutodeterminethenumberofpossibledesignsthatexitedofyouwerefacedwiththeproblemofaddingthree