• Tidak ada hasil yang ditemukan

Pacheco - stakeholder identification methods - 2012.pdf

N/A
N/A
Protected

Academic year: 2023

Membagikan "Pacheco - stakeholder identification methods - 2012.pdf"

Copied!
11
0
0

Teks penuh

(1)

ContentslistsavailableatSciVerseScienceDirect

The Journal of Systems and Software

j our na l h o me p a g e :w w w . e l s e v i e r . c o m / l o c a t e / j s s

A systematic literature review of stakeholder identification methods in requirements elicitation

Carla Pacheco, Ivan Garcia

PostgraduateDepartment,TechnologicalUniversityoftheMixtecRegion,Mexico

a r t i c l e i n f o

Articlehistory:

Received2August2011

Receivedinrevisedform27April2012 Accepted30April2012

Available online 8 May 2012

Keywords:

Systematicreview Requirementsengineering Stakeholderidentification Requirementselicitation Softwareengineering

a b s t r a c t

ThispaperpresentsasystematicreviewofrelevantpublishedstudiesrelatedtotopicsinRequirements Engineering,specifically,concerning stakeholderidentification methodsinrequirementselicitation, datedfrom1984to2011.Addressingfourspecificresearchquestions,thissystematicliteraturereview showsthefollowingevidencegatheredfromthesestudies:currentstatusofstakeholderidentification insoftwarerequirementelicitation,thebestpracticesrecommendedforitsperformance,consequences ofincorrectidentificationinrequirementsquality,and,aspectswhichneedtobeimproved.Ourfindings suggestthattheanalyzedapproachesstillhaveseriouslimitationsintermsofcoveringallaspectsof stakeholderidentificationasanimportantpartofrequirementselicitation.However,throughcorrectly identifyingandunderstandingthestakeholders,itispossibletodevelophighqualitysoftware.

© 2012 Elsevier Inc. All rights reserved.

1. Introduction

TheRequirements Engineering(RE) areaisan essential part ofanysoftwaredevelopmentprojectthatspecifies,analyzes,and definestheproductgoal,functionality,andlimitationsofthefinal product(IEEE,1998;Wiegers,2003;HofmannandLehner,2001).

Thefactthatsoftwarerequirementshaveasignificantimpacton finalsoftware productqualityimplies thatit isreasonably well documented(Liscomb,2003;Standish2009;IEEE,2004;SEI,2006).

UsuallyREcanbedescribedasacommonseriesofstagesinclud- ingelicitation,analysis,specification,validation,andmanagement (Pressman, 2005; Sommerville and Sawyer, 1997). In addition, threeofthemostimportantcategoriesofproblemsaffectingthe correctnessofsoftwarerequirementsaredefinedintheliterature:

gaining,comprehensionandvolatility(LoucopulosandKarakostas, 1995;SommervilleandSawyer,1997;KotonyaandSommerville, 2000).

Ourresearchis focusedonthefirstof thesestages,namely:

requirementselicitation.Requirementselicitationisrecognizedas oneofthemostcriticalactivitiesofsoftwaredevelopment(Davis et al.,2006);poorexecution of elicitationwillalmost certainly guaranteethatthefinalprojectisacompletefailure.Sinceproject failuresaresouncontrolled(Standish,2009),itisquitelikelythat improvinghow theindustry performs elicitation would have a

Correspondingauthor.

E-mailaddresses:[email protected](C.Pacheco), [email protected](I.Garcia).

dramaticeffectonthesuccessrecordoftheindustry(Hofmann andLehner,2001).Improvingrequirementselicitationrequiresus tofirstunderstandthestakeholderidentificationphase(Nuseibeh and Easterbrook, 2000). In the case of requirements elicitation activities – in which the problem to be solved is identified – the mostimportant thing is that thestakeholders be correctly identified(SEI,1992).Relationshipsandwaysofcommunicating betweenthedevelopmentteamandthecustomerareestablished atthistime(ISO,2004;Sommerville,2002).Despiteitsimportance, theidentificationofstakeholders,includingtheidentificationof their needs and expectations, is poorly achieved in software projects(Sommerville,2002;Pressman,2005).Oneprobablecause isthatthisprocessismistakenlyviewedasaself-evidenttaskin whichdirectusers,clientsandthedevelopmentteamaretheonly stakeholders.Itcouldalsobeduetothefactthattheidentification areacanbeeliminatedorsubstitutedbyopinionsorknowledge obtainedfromother moreaccessible sources ofinformation. In theshortterm,thiswouldcreatelessconflictsofinterestresulting fromdifferentpointsofview(Smith,2000).

Theoretical and empirical approaches are now beingunder- takenmoreoftentoinvestigateawideningrangeofphenomena inSoftwareEngineering(SE)specificallyinrequirementselicita- tionaspartofRE.Aseachapproachiscertainlylimitedinscope, researchersneedtobeabletorigorouslyandsystematicallylocate, assessandaggregateoutcomesfromallsignificantempiricaland theoreticalstudiesrelatedtoaparticulartopicofinterest,inorder toprovideanobjectivesummaryoftherelevantevidence.Thisneed hasbeenaddressedthroughtheapplicationprocessofaSystematic LiteratureReview(SLR).

0164-1212/$seefrontmatter© 2012 Elsevier Inc. All rights reserved.

http://dx.doi.org/10.1016/j.jss.2012.04.075

(2)

Inpreviouswork,weperformedanempiricalstudytoidentify howthestakeholderidentificationprocesscanaffectrequirements qualityand,asaconsequence,thedevelopedsoftwarequality.This study,presentedinPachecoandTovar,proposedacategorization ofstakeholderidentificationmethodsinrequirementselicitation.

Ourworkinthispaperfocusedonthreeissuesthathadnotbeen examinedinearlierwork.First,althoughstakeholderidentifica- tionmethodswereidentifiedweneededtodetermineandprovide evidenceabouteffectivepracticesrecommendedtousethem.Sec- ondly,oncethoseeffectivepracticeswereexposedweneededto determinetheconsequencesofincorrectlyperformingthestake- holderidentificationmethods;andthirdly,withthecollecteddata ofpreviousissues,wesummarized whataspectsof stakeholder identificationneededtobeimproved.

Consequently,inthispaper,wepresentanddiscussourexpe- riencesofapplying thesystematicliterature reviewin orderto gatherandevaluateavailableevidencepertainingtoStakeholder Identification(SI)inrequirementselicitation.

Thispaperis organizedas follows: Section2 presentsother approachesrelatedtoSLRinrequirementselicitation;Section3 describesthemethodusedforourSLR,reportingonthequalityof thepapersincludedinthissection;Section4reportsontheresults ofoursynthesisofidentifiedtopicsbasedonourfourresearchques- tions;Section5presentssomeofthelimitationsofthisstudy;in Section6,wegivesuggestionsforfurtherresearch;andfinallyin Section7,wepresentourconclusions.

2. Relatedwork

ThereisstillnosubstantialresearchrelatedtoSLRinrequire- mentsspecificationand elicitationtechniques,and aboveall,in stakeholderidentification,asseenbelow.

In 2006, Davis et al. (2006) reported a systematic review of empirical studies concerning the effectiveness of elicitation techniques, and the subsequent aggregation of empirical evi- dencegatheredfromthosestudies.Thesearethemostsignificant resultsthat wereobtained: (1) interviews,preferentially struc- tured,appeartobeoneofthemosteffectiveelicitationtechniques;

(2)manytechniquesoftencitedintheliterature,likecardsorting, rankingorthinkingaloud,tendtobelesseffectivethaninterviews;

(3)analystexperiencedoesnotappeartobearelevantfactor;and (4)thestudiesconductedhavenotfoundsignificantpositiveeffects fortheuseofintermediaterepresentationsduringelicitation.

ChengandAtleereviewedREresearchandidentified“future”

researchdirections suggestedbyemergingsoftware needs.This research examined technologies developed to address specific requirementstasks, suchas elicitation, modeling, and analysis.

Suchareviewenabledauthorstoidentifymatureareasofresearch, as well as areas that warrant furtherinvestigation. Next, they reviewed several strategies for performing and extending RE researchresults, tohelp delineate thescopeof future research directions(ChengandAtlee,2007).

Davisetal.(2006)proposedrecommendationsbasedonthepre- vioussystematicreview,whichwasupdatedandexpandedwith 13newempiricalstudiesandmorethan60newempiricalresults, topresentsomerecommendationsaboutthesituationsinwhich elicitationtechniquesareuseful(Diesteetal.,2008).

NicolásandTovalpresentedasystematicreviewoftheliterature relatedtothe generationoftextual requirementsspecifications from software engineering models. According to the results obtained,thebenefitsofbothlistsoftextualrequirements(usu- allywritteninnaturallanguage)andsoftwareengineeringmodels (usuallyspecifiedingraphicalform)–canbebroughttogetherby combiningthetwoapproachesinthespecificationofsystemand softwarerequirementsdocuments(NicolásandToval,2009).

Condori-Fernandezetal.describedanempiricalmappingstudy, whichwasdesignedtoidentifywhataspectsofsoftwarerequire- mentspecificationswereempiricallyevaluated,inwhichcontext, andbywhichresearchmethod.Onthebasisof46identifiedand categorizedprimarystudies,authorsfoundthatunderstandabil- itywasthemostcommonlyevaluatedaspectofSRS;experiments werethemostcommonlyusedresearchmethod,andthattheaca- demicenvironment waswhere mostempiricalevaluationtakes place(Condori-Fernandezetal.,2009).

DiesteandJuristopresentedtheresultsofasystematicreview of 564 empirical studies on elicitation techniques and aggre- gatedtheseresultstogatherempiricallygroundedevidence.They selectedandextracteddatafrom26ofthosepublications(contain- ing30empiricalstudies)toprovideasetofelicitationapplicability guidelinesbasedonthegatheredpiecesofknowledge.Theirgen- eralfindingisthatinterviewsarethemosteffectiveofallofthe testedelicitationtechniques(althoughtheyarepossiblylesseffi- cient insomedomains thanothertechniques,likeladderingor sortingtechniques).Likewise,theauthorsdonotrecommendthe useofintrospectivetechniques(i.e.,protocolanalysis)becausethey faredworsethanalloftheothertechniquesinallofthetested dimensions(effectiveness,efficiency,and completeness) (Dieste andJuristo,2011).

Insummary,thesestudiesonlycoversomeaspectsofrequire- ments elicitation; however none of them analyze stakeholder identificationevidence, despitethis being a crucial partwithin requirementselicitation.

3. Researchmethod

ASLRis“ameansofevaluatingandinterpretingallavailable researchrelevanttoaparticularresearchquestionortopicareaor phenomenonofinterest”(Kitchenham,2004).Theresearchpapers summarizedinthereviewarereferredtoasprimarystudies,while thereviewitselfisa secondarystudy.Theaccumulationofevi- dencethroughsecondarystudiescanbeveryvaluableinoffering newinsightsorinidentifyingwhereanissuemightbeclarifiedby additionalprimarystudies.

ASLRexaminesandinterpretsallavailableresearchrelevantto aparticularquestionortopicarea.Itaimstopresentanevaluation oftheliteraturerelativetoresearchingatopicbyusingarigorous andauditablemethodologysummary(Beechametal.,2007).

So,duetotheimpactofSEandREonsoftwarequality,asmen- tionedbrieflyinSection1,weconductedasystematicreviewtosee howSIisperformedandhowitcanbeimproved.

We followed guidelines derived from those used by medi- cal research, adapted and applied by Kitchenham (2004) and Kitchenhametal.(2004)toreflectthespecificproblemsofSEand REresearch(i.e.Beechametal.,2007;Breretonetal.,2007).

In accordance to Kitchenham (2004) and Kitchenham et al.

(2004),wetookthefollowingsteps.

3.1. Identifytheneedforasystematicliteraturereview

REisadisciplinethatarosewhenitbecameevidentthatthe qualityofrequirementsspecificationwasthekeyfactor inpre- venting,withtheleastpossiblecost,manyofthecausesleading tosoftwarefailure(Raghu,1995).Thus, effortsin this direction employedatanearlystageofaprojecthavegreatrepercussions, andarealsomoreprofitablethanothereffortscarriedoutafter- wards.The problemof the“software crisis”has,therefore, toa great degreeshiftedtorequirements.But,is there someaspect withintherequirementsareathatdeservestobegivenparticu- lartreatment?Basedonourpreviouslineofthinking,thisaspect shouldbeconnectedtooneoftheinitialactivitiesofRE;thatis,

(3)

thecaseofrequirementselicitation,anactivitywheretheproblem tobesolvedisdiscovered,andmoreimportantly,thestakeholders areidentified,therebyestablishingtherelationshipsandtheways ofcommunicationbetweenthedevelopmentteamandtheclient (IEEE,2004).

Softwareengineersneedtoidentify,characterize,andhandleall theviewpointsofthedifferenttypesofstakeholders(Kotonyaand Sommerville,2000).Thestakeholdersmayvaryfromoneproject toanother.Itis,therefore,alwaysnecessarytocarryoutanadap- tationassessmentofstakeholders’contributionsandtheirvested interestsinaproject(Raghavanetal.,1994).However,SI,aswellas identificationoftheirneedsandexpectations,ispoorlyconducted insoftwareprojects(SEI,2006),probablybecausethisprocessis mistakenlyseenasaself-evidenttaskwheredirectusers,clients, andthedevelopmentteamarethesolestakeholders.Itcouldalso beduetothefactthattheidentificationareacanbeeliminated orsubstitutedbyopinionsorknowledgeofothermoreaccessible sourcesofinformationthat,intheshortterm,producelessconflict ofinterest,asdifferentvisionsexistandmaycausedisagreement.

TheSIimpactonthequalityofsoftwarerequirementsisreflected inthesuccessachievedintheprojectaswellas,inthecurrent practicesusedtocarryoutthistask.

Asaconsequencetothegrowingnumberofstudiesinempirical andtheoreticalresearchinREitispertinenttoapplyasystematic approachtoassessingandaggregatingresearchoutcomesinorder toprovideabalancedandobjectivesummaryofresearchevidence.

ThereforeweneedtoapplytheSLRspecificallyinrequirements elicitation,toobtainresearch evidencefor SI methods ortech- niques.Also,wewouldshowwhataspectssoftwareengineersneed toimprovein requirementselicitation(Mitchell etal.,1997)in ordertoproducehighqualitysoftware.

3.2. Formulatereviewresearchquestions

Oursystematicapproachtoanalyzingpublishedstudiesenables ustoidentifyreliablywheretheliteraturepresentsthedifferent practicesdevelopedtocarryoutthistask.

Wesummarizedthisevidenceinordertoknowwhataspects needtobeadaptedtoimprovetheSIandwhatprocessisneces- sarytocarryoutanadaptationofstakeholders’assessment,their contributionsandvestedinterestsinaproject.Welookedatthe literaturetoanswertheseresearchquestions:

Question1:Whatmethodsortechniquesarecurrentlyusedtocarry outSIinRequirementsElicitation?

Question2:Whataretheeffectivepractices1recommendedforper- formingSI?

Question3:WhataretheconsequencesofincorrectSIonthequality ofSoftwareRequirements?

Question4:Whataspectsof SIarenecessaryto useasadvisable practices?

Fig.1givesanoverviewofhowourfourresearchquestionswork togethertogiveacomprehensiveviewofourtopic.

3.2.1. Searchterms

Fromourfourresearchquestions wederived thekeywords:

“Requirementselicitation,Stakeholderidentification,Method,Tech- nique,Effective practices”. Asearchstringwasconstructedusing relevanttermsbasedonresearchquestions.Alsowemadealist of synonyms ofeach ofthese keywords,as in theexample for

1Effectivepracticesareactivitiesthatpeoplewithrecognizedexpertiseinapar- ticularareahaveidentifiedfromexperienceasmakingsignificantcontributionsto projectsuccess(TchidiandZhen,2010).

RQ1, which contains keywords “stakeholderidentification” and

“requirementselicitation”:

Keywords((elicitation*ORobtaining*ORgaining*ORextract- ing* OR acquisition* OR discovery* OR capture*) AND (stakeholder*ORinterestedparty*ORpersoninvolved*)AND (identification*ORclassification*ORcategorization*ORrecog- nition*ORnaming*ORdetection*)).

WeexpandedthetermsusingWordNetVersion3.0(Princeton, 2010),andSoule’sdictionaryofEnglishsynonyms.Thelistofsearch termswasadaptedtomatcheachofourfourresearchquestions.

3.3. Searchingstrategy

The SLR process recommends searching several electronic sources(Kitchenham,2004;KitchenhamandCharters,2007),so weusedthefollowingsevenelectronicdatabases:

•ACMDigitalLibrary

•IEEEXplore

•SpringerVerlag

•GoogleScholar

•ScienceDirect

•Metapress

•WileyInterScience

In ordertodetermineifsimilarworkhadalready beenper- formedandlocatepotentiallyrelevantstudies,thesearchstrategy for the review was primarilydirected toward findings in pub- lishedpapers(journals,conferenceproceedings,technicalreports, or books) fromthecontent of the7 electronicdatabases men- tionedabove,althougheachidentifiedprimarysourcewaschecked forotherrelevantreferences.Weconductedtrialsearchesusinga numberofsearchstringsconstructedusingacombinationofkey- wordsandsynonymsmentionedinSection3.2.1.

3.4. Studiesselection

TheselectionofmaterialforourSLRwasbasedonthefollowing criteriaandprocedure.

3.4.1. Studiesselectioncriteria

Themain criterion forinclusion asa primarystudy wasthe presentationofempiricalorpracticaldatashowinghowtheSIis carriedout.AllthematerialusedinourSLRwasselectedbasedon thefollowinginclusioncriteria:

•Directlyansweranyoneormoreofourresearchquestionsor synonyms(seeSection3.2.1).

•Waspublishedinyears:19932–2011.

•Relatestoanypractitionerdirectlyproducingsoftware.

SincethefocusofthisreviewistheSIinRequirementsElicita- tion,weexcludedtexts:

•Intheformofslidepresentations.

•Workshops.

•Opinions,viewpointsoranecdotes.

•Toolswithoutempiricalevaluationoftheirapplication.

2Theyear1993waschosenasthebaselinebecausethefirstREsymposiumwas in1993.

(4)

Fig.1.Therelationshipbetweenourfourresearchquestions.

Table1

Qualityassessment.

Assessmentcriteria Score Responseoptionsforscore(fieldin Endnote)

Istheaimoftheresearch sufficientlyexplained?

Yes=1/moderately=0.5/no=0 Isthepresentedapproach

clearlyexplained?

Yes=1/moderately=0.5/no=0 Forapaper,whatisthe

acceptancequalityrate basedonthefindings?

Nofindings=0Over80%=1/under 20%=0/between=.5

% Enter%inqualityassessmentfieldin

Endnote

Thesecriteriawereappliedtostudiesinindustrialandacademic environments.3

3.4.2. Studiesselectionprocedure

Thepreliminaryselectionofprobableprimarystudieswasini- tiallybasedonreviewoftitle,abstract,andkeywords,although thissearchwasextendedtoincludeaconclusionssectioninthe caseswhere title, keywordsand abstract didnotprovide suffi- cientinformation.Afterthis, allselectedsourceswerereviewed againstadetailedsetofourinclusioncriteriaappliedoverallpub- licationstoobtaintheprimarystudies.Furthermore,weconducted secondarysearchesbasedonreferencesfoundinourprimarystud- ies.Allresearcherswerepromptedtorecordadditionalreferences forfollow-upworkontheprimarystudies‘results’form.

Inordertoavoidanystudyduplication;weexaminedallthe studiestofindrepeatedpublications,i.e.ifasimilarstudywerepub- lishedindifferentpublications,evenwithdifferentfirstauthors, onlythemostrecentorbroadeststudywasincludedinthereview (iftwostudieswerefoundtobeequallydatedandbroad,onestudy wasincluded).

3.5. Studyqualityassessment

Thestudyqualityassessmentcanbeusedtoguidetheinterpre- tationofthesynthesisfindingsandtodeterminethestrengthofthe elaboratedinferences(KitchenhamandCharters,2007).Thequal- ityofeachacceptedstudywasevaluatedaccordingtothecriteria showninTable1.

3 Specifically,empiricalstudiesbasedondirectevidenceorexperimentsinindus- trialcontexts,andtheoreticalorconceptualstudiesbasedonanunderstandingof thefieldfromexperienceandreferencetootherrelatedwork(i.e.academicsfind- ings)wereanalyzedaccordingtoourselectioncriteriatodeterminestudyinclusion orexclusion.

Withthefirstcriterionweassessediftheauthorsofthestudy clearlystatetheaimsandobjectivesoftheconductedresearch.

Thisquestioncouldbeansweredpositivelyforallthereviewed publications.Withthesecondcriterionweaskedifthestudypro- videsenoughinformation(eitherdirectlyorbyreferencingtothe relevantliterature)togivethepresent researchtheappropriate contextandbackground.Foralmostallpublications(87.5%)this wasansweredpositively.Thelastquestionallowsustoassessif theoutcomeoftheresearchwassufficientforourresearchpurpose.

Thisquestioncouldbeansweredpositivelyinalmostallpublica- tions(83.3%).Theappraisalmeasureswereestablishedbyagroup ofthreeexperienced softwareengineeringresearchersfromthe TechnologicalUniversityoftheMixtecRegionandvalidatedbyour independentreviewer.However,thescoringisaheuristiconlyto beusedasaguideandnostudywasrejectedonthebasisofitsqual- ityscore.Wenormalizedthedatafromthe47papers,combining thepercentageobtainedinthequalitycriterion(seeTable2).

3.6. Dataextractionstrategy

ThedataextractionprocesswasconductedusingtheEndnote version 9,todocument referencesforeach study.Accordingto Beechamet al.(2007), eachstudy usedtoanswertheresearch question(s)wasrecordedonaseparateresultsform,withtheaim ofidentifyingthetopicsemanatingfromthefindingsreportedin eachacceptedpaper.Inourcase,theseidentifiedtopicsgaveus thecategoriesreportedinourfindingsandresultssection.Wealso discoveredafterdataextractionthatwedidnothaveasignificant numberofpublicationsthatprovidestepsorguidelinestocarryout thestakeholderidentification(seeSection4).

3.7. Synthesisoftheextracteddata

Kitchenham’sguidelinesarenotentirelyclearaboutthenature ofthedataextractionprocess–howmuchcategorizationisdone duringdataextraction,andhowmuchisdoneinthedatasynthesis step.Weoptedfortrivialdataextractionresultinginalistofquotes thatwereonlyminimallyparaphrased.Wecategorizedtheseinthe earlypartsofthesynthesisstage.

InSection4wepresentfrequenciesofthenumberoftimeseach themeisidentifiedindifferentsources.Wegiveeachoccurrence thesameweight.Thefrequenciesmerelyreflecthowmanytimes agivenissueisidentifiedindifferentpapers,nothowimportantit maybe.

3.7.1. Documentretrieval

Oursearchesallowedustoobtainmorethan980references.

Analyzingthetitle and abstract,wecouldrejectapproximately

(5)

Table2

Qualityscoresofacceptedpapers.

Quality(scores) Total

Poor(<26%) Fair(26–45%) Good(46–65%) Verygood(66–85%) Excellent(>86%)

Numberofstudies 4 6 19 11 7 47

Percentageofpapers ∼8.5% ∼12.7 ∼40.4% ∼23.4% 14.9% 100%

Over78%ofpapersincludedinourliteraturereviewhavequalityscoresthataregoodtoexcellent.

735ofthese(seeSection3.4.2).Thenumberoffalsepositivesin theinitialset(papersthatmayhavebeenrelevantbutondetailed investigationturnedoutnottobeso)wasdisappointinglyhigh.We thenlookedat245papersinfulltoestablishafinallistof47papers.

AllthestepsinvolvedintheselectionprocessareshowninTable3.

Ourvalidationexercisesincluded:

(1)Inter-rateragreement:AccordingtoFleiss’Kappa(Fleiss,1971) theinter-agreementdenotestherebythedataextractioncon- sistency between the research studies when two or more researchersassesseachpaper.So,weraninter-rateragreement testsonthe245paperreferenceswefoundinourpreliminary search.Theprimaryresearchergroup,conformedbyexperts fromtheTechnologicalUniversityoftheMixtecRegionana- lyzedmeticulouslyeachoneofthesepapers(9papersappeared asunobtainable).Theprimaryresearchersaccepted50papers.

Ourindependentresearcheranalyzedrandomly28paperscho- senamongrejectedandacceptedpapers(approximatelyeach 8thstudyfromanalphabeticallistof245).A99.4%conformity wasrecordedwiththeoriginalassessments.Thispercentageof agreementallowsustohavecertaintyinouracceptanceand rejectiondecisions.

(2)Independentappraisal:Weconductedthisvalidationexercise onthe51acceptedstudies.Inthisexercisewehadahighper- centage ofagreementbetweentheprimary researchersand the independentexpert (99.8%), and any disagreementwas discussed. Therewasdisagreement on fourof theaccepted papers,sowerequestedtheopinionofasecondindependent researcher,whoagreedtorejectthesefourpapersafterhav- ingtakenintoaccounthoweachstudyansweredourresearch questions.Asaresultofthisexercise,47paperswereleftfor inclusion.

Table3

Papersreviewedandvalidated.

Selectionprocess #Papers Papersusedinvalidation Papersextractedfrom

electronicdatabases

<980 n/a Siftbasedontitleand

abstract

245 n/a

Papersfullversions available[245–9]

236 [28papersrandomlyselectedfrom thissetforvalidation1]

Papersaccepted(by primaryresearchers)

50

Papersrejectedby independentresearcher (validation1)

49 [1paperrejectedfromthe28 papersamplethatformedpartof the50acceptedpapers]

Papersaddedby independentresearcher (validation1)

51 [2papersacceptedoutofthe28 randomlyselectedpapersthat werepreviouslyrejectedby primaryresearchers]

Papersrejectedin validation2(47papers remaininourreview)

47 All51papersassessedand qualitativeformscompletedbyan independentresearcher4 rejected

4. Results

Atotalof47studiesdiscusstheSImethodsinRE.Citationsfor47 papersincludedinthissectionaregiveninnumericalformwitha bibliographyforfurtherreading.Itisveryimportanttomention that, duetotheunstructured sourcesand highlevelof hetero- geneityoftheanalyzedstudies,meta-analysistechniquessuchas aggregateddataorcouldnotbeapplied(wecannotcombinethem) (Kitchenham,2004;KitchenhamandCharters,2007).

Also,itisveryimportanttosaythatasthemajorityofthestudies analyzedarenotvalidated,wewereunabletoanalyzetheirimpact inrequirementselicitation.Priortopresentingresultsandanalysis foreach researchquestionwewillgivea shortoverviewofthe generalcharacteristicsofthestudies.

4.1. Overviewofthestudies 4.1.1. Researchmethod

The inspected publications were classified according to the applied research method. Our initialstrategy of categorization wassimpleandstraightforward:extractthementionedresearch method without interpreting the content study. Therefore, we definedthefollowingcategoriestoclassifythestudies:

•Empirical,i.e.findingsarebasedondirectevidenceorexperi- ment.

•Theoreticalorconceptualstudiesbasedonanunderstandingof thefieldfromexperienceandreferencetootherrelatedwork.

•Others.

Fig.2showsthatoutofthe47studies,90%areempirical,8%

theoreticaland,asmallnumberofstudies(2%)areeitherreviews

Fig.2.Researchmethodinouracceptedpapers.

(6)

Fig.3. Numberofpapersincludedinthereviewby5-yearintervals.

oftheliteratureorsecondarystudies,whereempiricalworkisre- examined.

4.1.2. Publicationyear

Thereviewedpaperswerepublishedbetween1993and2011.

Fig.3 shows that over 14 years (1993–2007) there is a recent increaseinpublishedpaperscoveringSI,specificallyfrom2000to 2007.Theincreasemaybeareflectionofagrowingawarenessof theimportanceofmotivationinREormayjustmatchageneral riseinpublishedpapersinSIinRE.Nevertheless,since2008the numberofSIstudieshasdecreasedpossiblyduetothefactthat researchonstakeholdersidentificationchangeditsfocustoother areassuchascharacterizationofstakeholdersinothermedialike socialnetworksaswecanseeinLim andBentley(2011),Costa andCunha(2010),Limetal.(2010a,b,2011),LimandFinkelstein (2011),Tsung-Tingetal.(2010),andWoolridgeandBailey(2011).

4.2. Stakeholderidentificationinrequirementselicitation

Byinvestigatingthefourresearchquestions,weaimtogaina broadpictureofwhattheliteratureisreportingonSIinrequire- mentselicitation.Wecollectedinformationaboutwhatmethods ortechniquesarecurrentlyusedtocarryouttheSIin Require- mentsElicitation(RQ1);whataretheeffectivepracticesinSI(RQ2);

whataretheconsequencesofincorrectSIonthequalityofSoftware Requirements(RQ3),and,whataspectsofSIarenecessarytouse asadvisablepractices(RQ4).Thefollowingsectionslookateachof ourfour-researchquestionsinmoredetail.

4.2.1. RQ1– methodsandtechniquestocarryoutthestakeholder identificationinRequirementsElicitation

FortypaperswereidentifiedinansweringResearchQuestion1 (RQ1):Whatmethodsortechniquesarecurrentlyusedtocarryoutthe SIinRequirementsElicitation?

Allthe studies analyzed gave us the impression that many attemptshavebeenmadetodefineandgivedetailedexplanations ofhowtheSIisdone.This,however,isnotthecase.Currently,stake- holderidentificationmethodsarefewandtheyarenotstructured aseachauthordescribestheprocessfromtheirviewpoint,lacking acommonframeworkofstudyandauniformdescription.Thehigh levelofheterogeneityofthestudiesanalyzeddoesnotallowusto presentaquantitativedataanalysis.

ThecurrentstatusofSIreferredtointhepresentpapershows differentinterpretationsofthescopeofthisprocess.Allofthesoft- wareinitiativesreferredtocontributetotheimprovementofthe softwareprocess,byimplementingasetofgoodindustrypractices forREthathavebeenidentified,acknowledged,anddisseminated.

However,theyhavenotexplainedhowtocarryouttheSI.

Table4

SIstudiescategories.

Categories Paperreferences Frequency(#

ofstudies) Studiesthatexclusively

describestakeholders.

BanAl-AniandEdwards (2004),BarberandJernigan (2000),Fuentes-Fernández etal.(2009),KasirunandSalim (2008),Lauesen(2002),and RobertsonandRobertson (1999).

6

Studiesfocusingonthe interactionbetween stakeholders.

AlexanderandRobertson (2004),ArguelloandCallan (2007),CoakesandElliman (1999),CoughlanandMacredie (2002),Coughlanetal.(2003), Damian(2007),Fassin(2009), GlinzandWieringa(2007), Hallingetal.(2003),InandRoy (2001),Kaiyaetal.(2005), Laportietal.(2009),Niuand Easterbrook(2008),Oshiro etal.(2003),Pahl(2004),Sharp etal.(1999),Smith(2000), StallingerandGrünbacher (2001),Söderholmetal.

(2007),StoneandSawyer (2006),PreissandWegmann (2001),Wong(2005),and Woolridgeetal.(2007).

23

Studiesthatincludean assessmentof stakeholders.

BallejosandMontagna(2006, 2008a),Davisonetal.(2006), GreerandRuhe(2004), Kulkarni(2008),McManus (2004),Mitchelletal.(1997), ParentandDeephouse(2007), Pouloudi(1997),Pouloudietal.

(2004),andYoungetal.(2001).

11

Someinitiativesprovide numerous examplesofwho canbe stakeholdersbyestablishing genericcategories intowhich they maybegrouped(seeSection4.2.1.1).Otherstudiesanalyzedare more ambitious.However, thestudiesmentioned in this paper are not standardized and consequently the SI is not standard- ized either. Also, not all of them cover the same aspects and thusarenotapplicabletothesamesituations.Thismakesitdif- ficulttoselectacorrectstakeholderidentificationmethodbecause somemethodsonlycharacterizethestakeholders,withoutassign- ing a stakeholder’srole in a specific project, nor analyzingthe stakeholderinteraction,norcoveringthehumanaspectsofstake- holder identification (see Section 4.2.1.2).Only a few methods include stakeholder assessment (see Section 4.2.1.3). Further- more, notall thestudiesanalyzed takeinto accountimportant aspects(Lewis,1991;Lloydetal.,2002)suchaswhenandhow we know that thestakeholders identified aresufficient for the project, and how all the information collected will be docu- mented.

In2007,PachecoandTovar(2007)identifiedthreeattributes relatedtothe‘issues’ofSIthatcanbestructuredintothreecat- egories: studies that exclusively describe stakeholders, studies focusingontheinteractionbetweenstakeholders,andstudiesthat includeanassessmentofstakeholders.Sowehavegroupedthe40 papersintothesethree categories.Section4.2.1.1givesthefirst categoryofstudiesthatlimitthemselvestoonlyproposingalistof possiblestakeholders.Section4.2.1.2presentsthesecondcategory ofstudiesthatnotonlyindicatewhothestakeholderscanbe,but alsostudytheirinteractions.Thethirdcategory,inSection4.2.1.3, dealswithstudiesthatincludeanassessmentofstakeholders(see Table4).Aswecansee,withintheREarea,therearenoguide- linesorproperstandardstohelpandguidesoftwareengineersin stakeholderidentification.

(7)

Table5

Listofpotentialstakeholders.

Typesofstakeholders Paperreferences Frequency(#of studies) Developmentteam,

clients/sponsors,and negotiators.

BanAl-AniandEdwards(2004) andKasirunandSalim(2008).

2

Customers,finalusers, developers,producers, teststaff,suppliers, marketingstaff,and maintenancestaff.

Fuentes-Fernándezetal.

(2009),BarberandJernigan (2000),Lauesen(2002),and RobertsonandRobertson (1999).

4

4.2.1.1. Studiesthatexclusivelydescribestakeholders. Accordingto PachecoandTovar(2007),thesepapersprovidealistofpotential stakeholdersfromwhich itispossibletodeterminewhichones arereallyrelevantandhoweach onemaybecontacted.Never- thelesstheyonlyprovideahelpfulguidetoestablishafinallistof stakeholders(seeTable5).

Whatmustnotbeoverlookedisthatstakeholderswillnormally havetocontributetheireffort,timeand/ormoney,andtheymust thereforeknowwhatbenefitscanbegainedinreturn.Potential stakeholdersmustthereforebecharacterizedbygatheringrelevant informationaboutthem.Thisinformationmayalsobeusefulfor evaluatingasetofidentifiedstakeholders,andforobtainingnew andmoreappropriateconfigurations(PachecoandTovar,2007).

Ingeneral,thesestudiescannotbeproperlyregardedasaniden- tificationofstakeholdersbecausetheyonlyprovideinformation thatfacilitatestheiridentification.Theydonotensurethatallthe necessarystakeholdersaredetected.

4.2.1.2. Studies focusing on the interaction between stakeholders.

PachecoandTovarproposedthatoncewehaveanideaofwhothe mainstakeholdersare,thebasicinteractionsbetweentheseactors shouldbeidentified.Thisenablesstakeholderstoclarifywhichpart oftheproblemfallswithineachone’sscope.“Interaction”involves communicating,readingasetofrulesorguidelines,searchingfor information,etc.(PachecoandTovar,2007).

TherangeofstudiesshowninTable6assignsstakeholderroles onthebasisofananalysisoftheinteractionsbetweendifferent stakeholdersandbetweenthestakeholderandthesystem.

4.2.1.3. Studiesthatincludeassessmentofstakeholders. Inthiscat- egorythemethodsshownsuggestthattheidentificationofpeople relatedtotheprojectwillbebasedontheirrelevancetotheproject

Table6

Wayinwhichtheinteractionbetweenstakeholdersandrelationshipsisestablished.

Manner Paperreferences Frequency(#of

studies) Usingacontextdiagram

toenablestakeholders toseewhatis happeninginthe system.

CoakesandElliman(1999), Damian(2007),Fassin(2009), Hallingetal.(2003),InandRoy (2001),Laportietal.(2009), NiuandEasterbrook(2008), PreissandWegmann(2001), Smith(2000),Stallingerand Grünbacher(2001),Söderholm etal.(2007),StoneandSawyer (2006),Wong(2005),and Woolridgeetal.(2007).

15

Establishingthe interaction,selectinga principalstakeholder andanalyzingall stakeholdersaround him(providers,etc.).

ArguelloandCallan(2007), CouhglanandMacredie(2002), Coughlanetal.(2003),Glinz andWieringa(2007),Kaiya etal.(2005),Oshiroetal.

(2003),Pahl(2004),andSharp etal.(1999).

8

Table7

Studiesthatincludeassessmentofstakeholders.

Categories Paperreferences Frequency(#of

studies) Priorityinterestsinthe

project.

BallejosandMontagna(2006, 2008a),Davisonetal.(2006), GreerandRuhe(2004), Kulkarni(2008),Mitchelletal.

(1997),ParentandDeephouse (2007),Pouloudi(1997), Pouloudietal.(2004).

9

Skillsandothersuitable attitudes.

McManus(2004),Youngetal.

(2001).

2

(priorityinterests),theirknowledgeandskills,andhavingasuit- ableattitudetowardtheprocess.SeeTable7.

Sofar,wehaveexplainedhoweachsoftwareprojectmayhave differenttypesofstakeholders,andhowselectingthemappropri- atelyhasastrongimpactonsoftwarerequirementsquality,and consequently,onthesuccessofthesoftwareprojectitself.

4.2.2. RQ2– effectivepracticesrecommendedforperformingSI SixteenpapersansweredResearchQuestion2(RQ2):Whatare theeffectivepracticesrecommendedforperformingSI?

Thereis generalagreement abouttheneed tofindeffective practicesrelating toSI inindustry.Weidentified thethreebest SIpracticesinthepapersanalyzed(seeTable8):

a.Identifyandconsultalllikelysourcesofrequirements:Thisprac- ticeisbasedontheknowledgethatstakeholdersshouldmeet thedemandsintermsofexperienceandexpertiseforeffective teamwork.So,itrequires(a)carefullyselectingteammembers whoareskilledintheapplicationdomainandREprocesses,(b) assigningexperienced,capableprojectmanagerstoRE,and(c) consultingdomainexpertsandstakeholdersatanearlystageof theprocesstoincreaseandvalidatetheteam’sknowledge.

b.Identifyuserclassesandtheircharacteristics:Thispracticeempha- sizestheneedforstakeholderidentification.Theremay,infact, bemanygroupsofcustomerswhousetheproduct,andthese canbeclassifiedintermsoffrequencyofuseoftheproduct,user characteristics,levelsofprivileges,orlevelsofskills.Sinceeach typeofproject(forexample,commercialapplications,integrated systems, web developments, etc.) requires different experts, properselectionof stakeholdersis recommended.Thisselec- tioninvolvesa previousassessment ofstakeholdersin terms ofrisk and cost,and alsotaking intoaccount standardtypes of communication betweenusers and developers. For exam- ple,communication in which developers can talkdirectly to

Table8 BestSIpractices.

BestSIpractices Paperreferences Frequency(#of studies) a.Identifyandconsult

alllikelysourcesof requirements.

BallejosandMontagna (2008b),CoakesandElliman (1999),HofmannandLehner (2001),ParentandDeephouse (2007),andRobertsonand Robertson(1999).

5

b.Identifyuserclasses andtheir

characteristics.

GlinzandWieringa(2007),Ko etal.(2007),Sharpetal.(1999), McManus(2004),Youngetal.

(2001)andWiegers(2003).

6

c.Identifyandconsult withthestakeholders ofthesystem.

CoughlanandMacredie(2002), Fuentes-Fernándezetal.

(2009),Kaiyaetal.(2005), Pouloudietal.(2004),and Wong(2005).

5

(8)

potentialusersismoreeffectivebecauseitavoidslossofinfor- mationduetotheuseofintermediaries.

c.Identifyandconsultwiththestakeholdersofthesystem:Thisprac- ticerecommendsmakingaveryspecificlistofstakeholdersat anearlystageoftheREprocess.It proposesamethodoffol- lowingguidelinesthatensureonlyappropriatestakeholdersare identifiedwithineachcategoryoftheproposedstakeholderclas- sifications.Itfurthersuggeststhatanexplicitlistofstakeholders bedrawnupandreasonsgivenwhytherequirementswillprob- ablybeimportant.

However,effectivepractices,orstandardssuchasCMMi,SWE- BOK,BABOK,andISO/IEC12207(SEI,2006;IEEE,2004;IIBA,2009;

ISO/IEC12207,2004)havethefollowinglimitation:theydonot explainhowtodefinetheentiresetofstakeholders.Furthermore, thisprocessisnotalwaysselfevident,andsoorganizationmustbe analyzedinordertoidentifyallpossiblestakeholders.Hence,the applicationofaSImethodsometimesbecomesindispensable.

4.2.3. RQ3–WhataretheconsequencesofincorrectSIonthe qualityofSoftwareRequirements?

OnlytwopaperswereidentifiedasansweringResearchQuestion 3(RQ3):WhataretheconsequencesofincorrectSIonthequalityof SoftwareRequirements?

Toanswerthis question,inthefirstplace, itis necessaryto considerwhatSoftwareRequirementSpecificationQuality(SRSQ) involves.TheIEEEStandard830(IEEE1998)givesasummaryofthe propertiesthatshouldideallybepartofsoftwarerequirementspec- ification:Correctness,Completeness,andConsistency.TheSIimpact onthequalityofsoftwarerequirementsisreflectedinthesuccess achievedinaproject,aswellas,fromthecurrentpracticeusedto carryoutthistask(PachecoandTovar,2007).

Anyidentificationprocess that mistakenly recognizessome- oneasastakeholderwillprobablyincluderequirements,whichdo notcorrespondtoanyrealneed(afeatureofcorrectnessofthe standard).Also,when theidentificationtaskfailstodetectpar- ticipantswhoareneeded forthesoftwareproject, requirement specificationsarenolongercompleteduetotheomissionofrel- evantrequirementsforprojectsuccess,andthiscouldgiveriseto inconsistentspecifications.Failingtoobtainthesepropertiescan createrisksthatcouldaffecttheproject.

Completeness,CorrectnessandConsistency intheSRSQcanbe ensured by applying proper SI and elicitation techniques such as: scenarios, case studies, etc. These three aspects are essen- tialtoobtainhigh qualityrequirements. The implications of SI onthequalityofrequirementsaresignificant.Forexample,the ISO/IEC25000:2503nSoftware QualityRequirementsand Eval- uation (ISO25010, 2011), and ISO/IEC 9126: Software Quality CharacteristicsandMetrics(ISO9126,2001)allmentionthatthe principlecharacteristicofsoftwarequalityrelatedtorequirements isfunctionality:theessentialpurposeof anyproductorservice (completeness,correctness,andsuitability),andthequalityofthe softwareusedissatisfaction:thecapabilityofthesoftwareproduct tosatisfyusersinaspecifiedcontextofuse.

Ontheotherhand,agoodSIcanprovidemanybenefits.Aproper selectionofstakeholdersimprovesthecoverageofrequirements, avoidsanoverlappingofrequirementsintheusercommunity,and allowsforamorerationalorganizationofrequirements.Inthisway, peoplegetinvolvedmoreeasily,andarelessreluctanttoimple- mentthesystemandgiveinformationrelatingtorequirements.

4.2.4. RQ4– SIadvisablepractices

Afterreadingandanalyze19papers,wecananswerResearch Question4(RQ4):Whataspectsofstakeholderidentificationarenec- essarytouseasadvisablepractices?

Table9

SIissuesthatmustbeimproved.

Issues Paperreferences Frequency(#of

studies) a. BallejosandMontagna(2008b),Mitchelletal.

(1997),Youngetal.(2001),McManus(2004), andVilleretal.(1999).

5

b. Al-SalemandSamaha(2007),Ballejosand Montagna(2006),CoughlanandMacredie (2002),Fassin(2009),InandRoy(2001),Ko etal.(2007),Sharpetal.(1999),Youngetal.

(2001),Kaiyaetal.(2004),andWong(2005).

10

c. Kulkarni(2008),ParentandDeephouse(2007), Pouloudietal.(2004),andStoneandSawyer (2006).

4

In requirements elicitation,the appropriate identification of stakeholdersisvitallyimportantasameansofunderstandingthe environmentinwhichthesoftwareprojectwillbedevelopedand operated,andalsotoidentifywhichstakeholderswillparticipate intheproject.Thisisakeyaspectintheprocessofobtainingthe expectedqualityrequirementsspecification,inthesensethatthey mustbeappropriate,complete,andfreeofcontradictions.

Aproperselectionofstakeholdersinordertoimprovethecov- erageofrequirements,avoidsanoverlappingofrequirementsin theusercommunity,andallowsforamorerationalorganization ofrequirements.Thismeansthatallstakeholdersneedtohavean appropriateknowledgeandnostakeholdercanbeomitted.

Inviewoftheaforementionedexplanation,thebenefitsofan adequateSIareevident.So,wecandefinetheSIprocessas:“A processthatcontributestotheidentificationofrelevantstakeholders inRE”(Conger,1994;Finkelstein,2000;Macaulay,1996;Zowghi andParyani,2003),and itmustcontainthefollowingadvisable practices:

a. Assignappropriaterolestostakeholdersthroughananalysisof skills,behavioringroupdynamicsandpersonalitytests;aspects thatwouldrendertheSIrepeatableandverifiable.However,it dependsonthestakeholders’availability.

b.Theestablishmentofconstructiveinteractionbetweenallstake- holdersduring the requirementsgathering process, and also betweenallstakeholdersandthesystemtoavoidconflictsand problems of communication arising from different points of view.

c.Classifyrequirementselicitedfromthestakeholdersaccording toanevaluationoftheirprioritiesinrelationtotheprojectgoal, in orderto definetheinteractions betweenthestakeholders themselves,andbetweenthestakeholdersandtheproject.This enablesustoverifywhethertheinitialprojectgoalhasbeen satisfied.

Table9showsthemostimportantcitedSIaspectsthatmust beimprovedaccordingtotheliterature:assigningofappropriate roles,establishmentofconstructiveinteraction,andclassification ofrequirementsaccordingtoanevaluationoftheirprioritiesinthe project.

5. Limitations

5.1. Completeness

Wehave conducteda verythoroughreviewof theliterature elicitingworkfrom38differentauthorsincludingsomesecondary studies(whereweusedthereferenceintheprimarystudytolead toanotherstudy).Wefoundthat47outof280studiespartially describestakeholderidentification(seeSection4).We notethat withtheincreasingamountofworkinthisareaduring1993–2007,

(9)

wecannotguaranteetohavecapturedalltheavailablematerialin thisarea.

Anotherareaofconcernisthatwedidnotconsiderstudiespub- lished ina non-Englishlanguage,thisis nota limitationof our approach,butareflectionofthelimitationsimposedonusbythe availableresearchinthisarea.

5.2. Publicationbias

Publication bias refers to the general problemthat positive researchoutcomesaremorelikelytobepublishedthannegatives ones(KitchenhamandCharters,2007).Weregardthis threatas moderate,sincetheresearchquestionsinthisreviewarenotgeared toward theperformance of a specificstakeholder identification methodortechniqueforthepurposeofacomparison.

Thestudiesincludedinthisreviewunderwentathoroughselec- tionprocessthatinvolvedseveralresearcherscross-checkingthe completenessof searches and validating thesuitability of each studyforinclusion.Thereforewedecidednottoincludegraylitera- ture(technicalreports,workinprogress,unpublishedornon-peer reviewedpublications).

However,asthereisasystematicbiasinthewaytheresearchis conductedinmanyoftheincludedstudies(oftenbasedonconve- niencesamples)wenotethatourresultsmaynotberepresentative ofallSImethods;forexamplewedidnottakeintoaccountnon- Englishlanguagestudies.So,wecannotgeneralizefromourresults.

5.3. Datasynthesis

DifferentareasinSoftwareEngineering,SoftwareRequirements andStakeholderIdentificationstudies,weregroupedinorderto identifytopicsthatanswerourresearchquestions.However,we mayhavelostsomeofthedetailsinchangesovertimebygrouping allthepaperstogetherbythemebecausetheyhaveahighlevelof heterogeneity.Also,welackdatesofpublicationintheperiodof 1984–2011.Whenweaggregatedourtopicsthereportedfrequen- ciesweretreatedwithmaximumcaution.

6. Furtherresearch

Thisreviewhasraisedmanyissuesthatwouldbenefitfromfur- therresearch.Forexample,instakeholderidentificationweneed thefollowing:

a.Totakeintoaccounttheimpactofpersonalitytypesandthe rolestheymayplay,becausethisaspectcouldbearepeatable characterofSI.

b.Totakeintoaccounttheprojecttypetobedevelopedbecause notallprojectsneedthesamestakeholderstoobtaintheproject requirements.

c.To assess stakeholders in terms of their characteristics, the knowledgeneeded,andtheirinfluenceonaprojecttodetermine thepriorityoftheirrequirements.

d.Touseschemestocharacterizeandevaluateappropriaterela- tionshipsbetweenallstakeholders.Forexample,wecoulduse thisresearchquestion:Doesthemethodanalyzedcontainlabels suchas,“onepersonisinchargeof”,“thispersonisanassistant to”,“he/sheiscrucialfor”,“he/sheprovidestheinformationfor”?

Alltheseissuescanbeusefulinordertoimprovestakeholder identification,andconsequentlyrequirementselicitation,which wouldinturnimprovesoftwarequality.

In addition,furtherworkis needed todevelop a methodor modeltocarryoutSIintheREarea.

7. Conclusions

Thispaperpresentsasystematicliteraturereviewthatinves- tigateshowstakeholderidentificationinrequirementselicitation is carried out. Theaim isto identifyand characterize different approachestoprovideacomprehensiveoutlineanddiscussionof methods,standards,and techniquesusedinRequirementsEngi- neering,specificallyinrequirementselicitation.

Intheareaofrequirementselicitationitiscriticaltodescribe thestakeholderidentificationprocessinordertoprovidecorrect, consistent, and complete requirements specification. However, oneofourfindingssuggeststhatduring2000–2007,anincreased interest for developing methods in SI existed, as compared to previousand posterior years.For example, wehave foundthat from 2008 until now, stakeholder identification changed focus tootherareassuchascharacterizationof stakeholdersin other media like social networks as we can see in Lim and Bentley (2011),Costaand Cunha(2010),Limet al.(2010a,b,2011),Lim andFinkelstein(2011),Tsung-Tingetal.(2010),andWoolridgeand Bailey(2011).

Also,wecanseethatstakeholderidentificationintherequire- mentselicitationphasehasreceivedverylittleattentionfromthe differentexistinginitiativesinsoftwaredevelopment,forexample inCMMi,SWEBOK,BABOK,andISO/IEC12207.Alltheseinitiatives recognizetheexistenceofdifferenttypesofstakeholdersinthe REarea.However,theyonlysuggestexamplesandcategoriesof stakeholders,anddonotprovidepracticesorguidelinestohelpand guidesoftwareengineerstoidentifystakeholders(whoneedtobe identifiedineachprojectasanindispensablepartofrequirements elicitation).

Despitethefactthatsuccessandqualityinsoftwareproducts depends to a great extent on requirements specification qual- ity,onlytwopaperstakestandardsintoaccount(ISO9126,2001;

ISO25010,2011).

Alltheanalyzedpapersconfirmthevarietyofexistingstake- holdersinvolvedinsoftwaredevelopment,eachhavingdifferent prioritiesandinterests.However,inrequirementselicitation,all ofthesestudiestakeSIforgrantedanddonotgobeyondindicat- ing“who”thestakeholdersmaybe.Softwareengineersneedto identify,characterize,andhandlealltheviewpointsofthediffer- enttypesofstakeholdersspecificallyintherequirementselicitation phase.

The SLR cites three advisable practices for improvement of thestakeholderidentificationprocess:“Forthestakeholderrole assignation they must be subjected to personality tests” (this dependingonthefreetimeofeachbusystakeholder),“theestab- lishment of constructive interaction between all stakeholders duringrequirementselicitation”(toavoidconflictsandproblems of communication), and “classification of requirements elicited according toan evaluation of theirpriorities in relationto the projectgoal”(toverifywhethertheinitialprojectgoalhasbeen satisfied).However,only19papersmentionsomeaspectsofthese issues.

Asafinalconclusion,theobtainedfindings,includinganexam- ination of theshortcomings found in this systematic literature review,providestrongevidencetoencouragefurtherresearchin thedevelopmentofanewmethodologytoadequatelyperformSI.In addition,weproposethedevelopmentofaguidethatrecommends theuseofaspecificmethodforstakeholderidentificationbasedon theparticularcharacteristicsoftheprojecttobedeveloped.

References

Alexander,I.,Robertson,S.,2004.Understandingprojectsociologybymodeling stakeholders.IEEESoftware21(1),23–27.

Referensi

Dokumen terkait

FK Universitas Tarumanegara Jakarta, http://journal.fkm.ui.ac.id/kesmas/article/view/347/346 diakses tanggal 8 desember 2018 Moyad, A.M., 2004, Fad Diets and Obesity - Part I:

Politics Quarterly, Journal of the Faculty of Law and Political Science 2022, 52 1: 153-173, DOI: 10.22059/JPQ.2022.272537.1007360 Research Paper The Political Thought of