onlinelotus

AnalysisandDesignforProcessSupportSystemsusingGoal-orientedBusinessProcessModellingByDeniseDowns,DrKenLunnSchoolofComputing&Mathematics,UniversityofHuddersfield,EnglandSummaryThispaperfocusesontheideasthatitisimportanttostartwithabusinessprocessmodelwhichfullyarticulatesthevisionandgoalsforthenewprocess(atwhicheverleveltheprocessiswithintheorganisation)whenproducingsoftware.
Furthermore,thesegoalsneedtobeauditablethroughtheprocesstoensurethedevelopedsystemassistsattainmentoftheprocessanditsgoals.
AsmallteamatHuddersfieldUniversityaredevelopingamethodologywhichcommenceswithvisionandgoalsofthebusinessprocess,requirescaptureofthenewprocessbutfacilitatesdevelopmentofprocesssupportsystemsaswellasstandardtransactionalsystemsandensurestherequirementsoftheprocessanditsgoalsaremetandtraceable.
Whileattheearlystagesthemethodologyintendstobuildoncurrentmethodsandtechniquesandatthisstagehasmainlyhighlightedissueswithcurrentapproaches.
Transactionalsystemshavedominatedtheliteratureintermsofdesignandsoftwaredevelopment.
Thisisaviewreinforcedbytheusecaseapproachtosoftwarerequirements[Rosenburg1999],wherethesystemisseenasasetoflargelydistinctfunctionalgroups.
Thisapproachhasledtoastyleofanalysisanddesignthatleavesalotofbusinessissuesoutsideofthesystem,andimplicitintheanalysis.
Processsupportsystemsarebeingperceivedasincreasinglyimportanttoanumberofcompanies.
Thisreflectsagrowingmaturityinorganisations,whicharekeentoautomatemoreoftheiractivities.
Applicationareas,suchascallcentres,canachievesignificantproductivityandqualityimprovementsbyutilisingprocesssupportandworkflowsystemstoimproveinformationflow,speedupresponse,andreduceproblemsindealingwithcustomers.
Customerrelationshipmanagementisrequiringgreaterunderstandingoftheprocessgoalsandvalues,andaconsolidationfromtransactionalsystemstoprovideforthebusinessprocessofmanagingcustomerswiththecaptureoftherelevantprocessdata.
Inessencethesetypesofactivityarebestcapturedbybusinessprocessmodels.
Workflowapproacheshavebeenreportedintheliterature,buttheyseemorthogonaltomuchoftheobject-orientedanalysisanddesignapproachesthataredocumented.
Theauthors'experienceinanalysinganddesigningprocesssupportsystemshitproblemswhenadvicewassoughtfromexpertsinothermethods.
Structuredmethodsexpertswantedtoconstructdataflowdiagramsthatdidnotadequatelyreflectprocessflowsinthebusiness.
Object-orientedexpertswantedtolookforusecasesandobjectrepresentations,withouttoocloseaconsiderationoftheprocessflow;indeed,modellingprocessflowinobjectsbegantolookover-complicated.
Noneofthemstartwiththeprocessgoals,visionandvaluesandprovideatraceableauditofensuringthesearemetthroughthedesign.
Whenthesoftwaredesignreflectsbusinessprocessestheresultingsoftwareseemstobemoreintuitivetousersandthuseasiertouse.
Further,wecontendthatfullacknowledgementofgoal-drivenbusinessprocessesleadstoamorecomprehensivedesignthatbettermeetsuserrequirements.
Howeverthetransitionfrombusinessprocessmodelstosoftwaredesignisnotatrivialone.
TheonlymethodwehavecomeacrossthatapproachesthisadequatelyistheSelectPerspective[Allen1998];thisapproachlooksatusecaseidentificationfromfullyanalysedprocessesusinganotationsimilartoIDEF;alasthismethodhasnotbeenfullydocumentedintheliterature.
Inthispaper,wedescribeourapproachtoanalysinganddesigningaprocesssupportsystem.
Webeginwithastatementofthevisionandgoalsofthebusiness,andananalysisandassessmentofthecurrentbusinessprocesses.
Wethenproducearevisedbusinessprocessdescription,andthenseekanobjectdecompositionandsetofuserinterfacestosupportthebusinessprocesses.
Wehaveinfactmissedoutanexplicitusecasestep,andwedonotseethisasaprobleminthisinstance,asdiscussedlater.
1BackgroundTheSchoolofComputingandMathematics'PlacementUnitundertookaprojecttoencapsulatetheactivitiescarriedoutattheSchoolofComputingandMathematics'PlacementUnit.
Theproject,MaPPiT(MappingthePlacementProcesswithInformationTechnology),hasresultedinthedevelopmentofacompleteElectronicProcessSupportsystem,designedanddevelopedusingLotusNotes,withsubsequentWeb-enabledsupportdevelopedinLotusDomino.
TheintentionwasasmuchtoencapsulateandcommunicatebestbusinesspracticeaswellastodevelopanITsystem.
Thesystemisnowoperationalandsupportsapproximately200placementstudentsayear,andbothprocessesandsystemarenowadoptedbyanumberofotheruniversities.
Thesystemsupportsthetrackingofstudentsandworkplacements.
Itprovidessupporttostudents,potentialemployers,academicstaffandsupportstaffthroughthewholeprocessofapplyingfor,achievingandundertakingaworkplacement.
ThisincludesmanagementofonlineCV'sthroughtoacceptanceofwork-basedassignmentsthatareusedaspartoftheassessmentofstudentsonplacement.
2AnalysisandDesignApproachWesplittheanalysisanddesignintotwophases.
Thefirstphaseistocomprehensivelymapoutthedesiredbusinessprocesses,defininggoalsandvision,andidentifyingissues.
Thisiscomprehensivelyrefinedthroughthefollowingstages:ArticulatedGoal,Vision,CurrentIssues/ProblemswithAs-IsProcessIdentifiedCore/KeyProcesses,ManagementProcesses&SupportProcessesCapturetasksincurrentprocessoncardsWorkinggroupsputtasksincohesivegroupsProcessNamesforgroupsdecidedProcessDesignedbyworkgroupsCheckwithoriginaltasksforcompletenessDocumentBusinessProcessDesignswithroles,triggersTheDevelopmentPhaseisbrokendownintothefollowingstages:IdentifyobjectsIdentifyProcessTrackingObjectsIdentifyAttributes–dataandstatusIdentifyBusinessPatternObjectslikeToDoProduceobjectdiagramforrelationshipsProduceUserInteractionMatrixforeachobjectusingRolesfromBPD&PFDOrganiseUserScreens(referredtoasuseragenda's)toreflectProcess2.
1Goaloftheprocess(es)Webeginouranalysisbydefiningthevisionandgoalsoftheoverallprocess.
Thecollectionofgoalsisacomplexactivity.
Inevitablythereareconflictingviewsonwhataretheappropriategoalsofaprocess,dependingontheviewsofstakeholders.
However,weviewitasessentialthatthereisagreementontheoverallgoalofthebusinessprocess(es)beforecommencingwithanalysis.
OuragreedgoalfortheMaPPiTsystemis:"Tosecure,developandmonitorrichlearningexperiences(onplacement)thatbuildonstudents'currentskillsandknowledge,inlinewiththeircareeraspirations;enhancetheiremployabilitythroughexperienceintheworkplace;andincreasetheirskillsandknowledge,subsequentlyenablinghigherlevelsofachievement.
"2.
2BusinessProcessModellingThebusinessprocessmodeldescribestheactivitiesofplacingstudentsintermsofaseriesofinter-linkedbusinessprocesses.
Itisastaticmodeluserscantailortotheirownrequirements.
Ittriestosatisfyanumberofaims,including:Todocumenttheplacementprocesses,followingareviewoftheirefficency,toensurethatnecessaryprocessesareincludedandnothingisomitted.
Toaddressissueswhichhademergedfromtheanalysisandotherprojects,eg.
someplacementunitsoperateinanunprofessionalway:missingdeadlines,notmeetingrequests.
.
Tofacilitateamovefromaplacementactivityrunpredominantlybyacademicstafftooneinvolvingmoreadministrativestaff;atHuddersfieldtherearetwofull-timeadministrativeplacementstaffintheunit;thisallowsustobemoreprofessionalandresponsivebutrequiresgreaterco-ordinationandattentiontocommunication.
Theseaimsgiveanunderstandingofwhatthemodelisabout,andguidethedefinitionoftheprocesses.
Thefocusisverymuchanoperationalone:theprojectteaminvestigatedwhatisdone,why,howitcanbeimproved,whatthestudentsthink,whatcompaniesthink,etc.
,andlookedtodefinegoodpracticeinthemodel.
Aspartoftheoperationalfocus,theteamexaminedwhatmaterialsareusedtopreparestudentsforplacements,thetraininggivenandtheskillsdeveloped.
Issuesthisraisedwerehowdoweauditourgoalstoensuretheyhavebeenaddressedintheresultingsystem2.
2.
1TheProcessesandSymbolsoftheModelTheprocessesofthemodelaregroupedintothreetypes:Core,ManagementandSupport.
Theunderlyingrationalebehindthesetypesis:CoreProcessesThesearekeytothebusinessunitandrepresentthecoreactivities.
TheymustbeundertakenwithinthePlacementUnitManagementProcessesIngeneralthesecanbemanagedbyindividualsnotinvolvedonadaytodaybasiswiththeplacementteamandrepresentreviewandmonitoring/directingprocesses.
TheymustinvolverepresentativesfromtheplacementteamSupportProcessesThesecangenerallybeoutsourcedtoothersupportinggroups.
Smallplacementteams,whodonothavetheresourcesforthefullmodel,may'passon'responsibilitytootherpersonnelwithintheschool/department(e.
g.
careers,personaltutors,pathwayleaders,commercialactivities/marketingetc.
).
2.
2.
2SymbolsUsedinTheModelThemodelswedevelopedusedanotationsimilartoprocessflowdiagrams(PFD)inIDEF3,shownasaflowofactivities(seeFigure1).
However,becauseweknewthenextstagewastodevelopasoftwareapplicationtosupporttherunningoftheseprocessflows,wepaidparticularattentiontothetriggers(seeFigure2)whicheitherstartedtheprocessorwererequiredtokeepitrunning–eachwaitstatehadatriggeridentifiedforthenextstage.
Wealsohadatimetriggerforthosetaskswewishedtokeepontimeforqualitycustomerservice,butweredependantonexternaltriggers(eg.
ReceivingaCVfromastudent).
Ineffect,iftheprocessflowdidnotprogressonit'sownnaturally(e.
g.
anemployerdidnotcontactustoorganiseinterviewsafterreceivingCV's),thetimetriggerwouldensurewebuiltintothesoftwaretheabilitytomanagethesethreads.
Oneachactivitywithinaprocessasuggestionismade,throughalabel,oftherolefelttobeappropriatetocarryoutthatactivity.
Activitiesaremarkedasessential,desirableoroptional.
Theteambelievesthatallthosewhicharemarkedasessentialshouldbecarriedoutbyanyplacementunit;desirableactivitiesarethosewhicharefelttobringbenefits,andoptionalactivitiesareforlargerplacementunitstoconsider.
Thisperspectivealsoprovidesaprioritysystemwhenconsideringwhichprocessestoadopt/review.
Figure1-notationforanactivityTrigger,aneventwhichisscheduledoranitemwhichcomesintothePlacementUnite.
g.
arequest.
Allprocesseshaveatriggeringevent.
TriggernameScheduledtrigger,otherwiseblankFigure2-notationforatriggerTheresultofourbusinessprocessdesignwasapproximately50processflowdiagrams,asillustratedinFigure3.
Manyoftheseweremuchmorecomplicatedwithbranchingandjoins.
Manyofthetaskswerepurepeopletasks,manywerepeopletasksthatrequiredtheITsupportsystem,veryfewwereITonlybecauseofthetypeofprocessweweremodelling–theplacementofstudentsintoindustryfortheir12monthworkexperience.
Whileverysimilartoarecruitmentprocesswealsohadtrainingandmonitoringoftheplacement.
The*IDActivitynameActivityconventionLetterdenotingthoseresponsibleforperformingtheaction:S=StudentVT=VisitingTutorPU=PlacementUnitstaffR=RecruitmentstaffDept.
=DepartmentT=TutorCL=CourseleaderCA=CommercialactivitiespersonnelUniquenumericalidentifier(LEVEL2only)Activitytoberepeated,otherwiseblankShadedmeansthisprocesshasbeenimplementedintotheEPSS.
Numberedpointsprovidefurtherinformation.
factthattherewerenosubstantialIT-onlytasks,wasafactorwhichmadethetransitiontoinformationsystemmoredifficult.
ReviewliteratureandusageCAActionplanactivitiesfornextperiod2CAReviewnumbersofandtypesofrequests1CAScheduledEnsureactioned-ifnot,decidealternativeactionForwardtoappropriateperson4PUFollow-uptoensureactioned5PULogdetails3PURequestEnterdetailsinsystemE.
g.
projects,consultancyFigure3-typicalbusinessprocessflows2.
3Designandconstructionoftheinformationsystem.
ThedeliveryplatformweuseisLotusNotes.
LotusNoteshasanobjectdatabasewithhierarchicalrelationships.
Collectionclasses(views)canbecreatedfromsimplequeriesonthedatabase.
Eachobjectisassociatedwithaformwhichprovidestheuserinterfacetotheobjectdataaswellasthevisibilitytothedesignerofthedataitemsintheobject.
OurapproachwastofirsttrawlthroughthePFD'Sandidentifyallourobjects.
Thisrepresentedacomprehensivedomainmodel.
Whatthisdidnotinitiallyrevealwastheobjectswerequiredtomanageourprocessthreads.
Thiswasthenexttrawlandweintegratedtheseintoourobjectmodelsowecoulddocumenttherelationshipwithourrealworldobjects.
Experiencealsotoldusthatweneededsomepatterntypeobjectslike'ToDo'and'ActivityReports'totrackandauditproblemsorspecialcaseswehadtodealwithwithintheprocess.
Theseweredifficulttodocumentontheobjectmodelasgenerallytheyweregoingtobemadeavailabletomostobjectsinthesystem.
ThisavailabilitywasdocumentedinaUserInteractionMatrix(e.
g.
Figure4).
Thenextphasewastoidentifyalltheattributesforeachofourobjects.
Someofthiswasfromexperiencee.
g.
Companyobviouslyrequiresname,address,telNo,butsomewerestatustypeattributes(placed/unplaced;monthtocontact)detectedfromthePFD's.
Thenextissuewasthatasourflowdiagramsweredecomposedtothethirdleveltoidentifythetasks,soanumberofdifferentdiagramsaffectedthefunctionalityofanumberofobjects.
Designingwhatactionsshouldbevisibletotheuseratwhatstageoftheprocessandwhatthatactionshoulddoifselectedwasproblematic.
ThesolutionwasaUserInteractionMatrix(e.
g.
Figure4).
Thisallowsustoidentifywhatmethods(viabuttons/actionsinLotusNotes)wereavailabletowhichusergroups(roles)interactingwiththeobjectandwhen.
ActionsCloseSaveJobDescriptionInterviewCorrespondenceCompleteActionDetail:CloseFormSavedetailscloseFormCreatenewJobdescriptiondetails-Updateformwithdate-Addinterviewees-createinterviewdetailsform-Schedulefollow-upcontactPromptforstandardlettertosendCreateletterwithcontactsdetailsSaveletterandprintPromptforoutcomeSetstatustosuccessfulorunsuccessfulVisibilityAllPUAdmin,PUAdmin,PUstaffPUStaffPUStaffPTutorsPTutorsStageNewPPPPDiaryreminderPPPPWaitingtoadvertisePPPAdvertisedPPPPInterviewsPPPPPSuccessfulPUnsuccessfulPFigure4-UserInteractionMatrixTheNotesInterfacehasit'sownparticularstyle.
ItpresentsaleftNavigationbartotheuserwithaseriesofbuttonsoricons.
Thesechangetheselectionintothedatabaseandtheresultingobjectsaredisplayedinadefinablewaytotheright.
EachKeyHighlevelProcesshasit'sownNavigatorbarandthebuttonsrepresenttheKeywaitpositionsintheprocessthatrequiremanaging.
WeaddedacoupleofextraNavigatorsforAdministration(ofthesoftware),Reportsandonewhichmirrorsthewebscreenswemakeavailabletostudents–Opportunities.
ThebuttonsontheNavigatorarepresentedintheordertheprocessruns,orsometimesimportantwaitstatesarehighlightedorthedefaultwhenthenavigatorisselected(e.
g.
openstudentactivityreportswhichrepresentproblemstobemanagedarefirstonallstudentscreens).
Emailsareusedfortriggerswhichdonotoccurveryofteneg.
astudentcourseworkarrives.
3DiscussionTheapproachherehasusedgoal-orientedbusinessprocessesasthedomainmodellingtoolforrequirementsanalysis.
Thisdiffersfrommanyapproaches.
TheUnifiedProcess[Jacobson1999]includesabusinessmodellingstep,butisnotprescriptiveontheapproachandimplicitlyassumesabusinessusecaseapproach[Jacobson1995]thatdiffersconsiderably.
OurapproachisakintotheSelectPerspective[Allen1999].
Wherewehavedifferedfromcontemporaryobject-orientedapproachesisinleavingouttheusecasestep[Rosenberg1999].
Ourusecasesareinfactautomated(orsemi-automated)processsteps.
Skippingtheexplicitusecasestepdidnotseemtocauseusanyparticularproblems,thoughelementssuchasrolesareembeddedinprocessstepsandareequivalenttoactorsinausecasediagram.
Thearchitectureofthetargetenvironment(LotusNotes)didlenditselftotheanalysisapproach.
WhenwetriedamoretraditionalOOapproachtodefiningthesystemidentifyingusersandusecases,wegotboggeddowninalotofdetailandcomplexity.
Atthisstage,wecannotbetooassertiveaboutwhetherthebusinessmodellingapproachissuperior,orwhetheritjustfitourparticularapplicationdomainandtargetsystemarchitecture.
Thesystemhasbeenremarkablysuccessful,andmeetsthegoalswesetit.
Itssuccesslies,webelieve,initbeingbasedonasolidunderstandingofthebusinessprocesses,andthegoalsofthosebusinessprocesses.
Wealsoweresystematicinfittingthefunctionalityofthesupportsystemtothebusinessprocess.
Also,alongwiththesystem,comesacleardefinitionofbestpractice,asdefinedbytheprocessmap.
InthePrismresearchgroupattheUniversityofHuddersfield,weareinvestigatingavarietyofapproachestorequirementsgatheringandsystemsanalysis.
Ourexpertisecomesfromabroadbackground,coveringsoftsystemsmethods,structuredsystemsmethods,objectorientedmethodsandbusinessprocessmethods.
Wearebuildingabroadpictureofmodellingapproachestosupportanalysisanddesign.
Bylookingatproblemsfromdifferentperspectives,wehopetouncoversomeunderlyingprinciplescommontoallmethods.
4References[Allen1998]:,Allen,P.
,Frost,S.
,Component-BasedDevelopmentforEnterpriseSystems:ApplyingtheSELECTPerspective,,ISBN0521649994[Bennett1999]:Object-OrientedSystemsAnalysisandDesign,Bennett,S.
,McRobb,S.
,Farmer,S,,1999,ISBN0077094972[Jacobson1995]:TheObjectAdvantage,Jacobson,I.
,,1995,ISBN0201422891,[Jacobson1999]:TheUnifiedSoftwareDevelopmentProcess,Jacobson,I.
,Booch,G.
,Rumbaugh,J.
,,1999,ISBN0201571692[Rosenburg1999]:,Rosenburg,D.
,Scott,K.
,UseCaseDrivenObjectModelingwithUML,1999,ISBN0201432897