Language Data The Other Data of Six Sigma.docx

Language Data The Other Data of Six Sigma.docx

《Language Data The Other Data of Six Sigma.docx》由会员分享，可在线阅读，更多相关《Language Data The Other Data of Six Sigma.docx（11页珍藏版）》请在冰豆网上搜索。

LanguageDataTheOtherDataofSixSigma

Language Data:

The‘OtherData’ofSixSigma,Part1of2:

NatureofLanguageData

DavidL.HallowellFebruary26,2010

Part1ofthisarticlefocusesonhelpingpractitionersunderstandtherolelanguagedataplaysinSixSigmaworkandhowtheycanbenefitfromunderstandingitsnature,includingthetypesandrefinementofthedataaswellashowtogatheritinthemosteffectiveway.

Part2–UsingLanguageDataisaboutthetoolsusedforprocessinglanguagedata.

ThereisakindofdatathatgetslesspressthanthenumberssovisibleinalmostallSixSigmawork –languagedata.Thefactthatitreceiveslessexposuredoesn’tmeanitislessworthyofexploration.

Isn’t SixSigmaAbouttheNumbers?

EveryoneinvolvedinSixSigma,fromleadersandchampionsthroughbeltsandteammembers,learnstheimportanceoffact-basedcommunicationtravelingintwodirections–input（listening,gatheringfactsand,distilling）andoutput（reporting,convincingandmotivating）.SixSigmahasdonesomuchtoimprovethewayquantitativeandgraphicaltoolsstrengthencommunication,itiseasytothinkthatthenumbersmustbethekeytosuccess.Inpractice,though,thenumberswithoutthestorywillnotdriveallthelearningandreportingnecessaryoverthecourseofmostDMAIC（Define,Measure,Analyze,Improve,Control）andDFSS（DesignforSixSigma）projects.Thisisespeciallytrueasthenumberofpeopleinvolvedincreases（Figure1）.

Figure1:

PointsofContactIncreaseCommunicationRisk

Afewpeopleworkingcloselytogetheronthesameprojectcanmanagecommunicationratherinformallyamongthemselves.Asthenumberofpointsofcontactgrowsthroughthesizeoftheteam,stakeholders,customers,suppliers,etc.,theneedforadditionallinesofcommunicationalsogrows.Inatypicalproject,thevariedperspectives,motivationsandlevelsoffocusacrosstheselinesmakeitclearthatcommunicationcannotbelefttoinformalmeans.

Afundamentaldrivetobetterunderstandlanguagedataisbuiltintotheproblem-solvingprocessthatisthefoundationoftheDFSSandDMAICmethodologies.Duringprojectselection,evenbeforegettingtothenumbers,onemustformulatetheproblemoropportunityusinginformationfromcustomers,stakeholdersandthetargetenvironment.Thatinformationislargelyintheformoflanguagedata.

Figure2illustratestheuseofdatatodriveprogressinproblemsolving.ThisbuildsonJiroKawakita’s‘W’modelandShojiShiba’s‘WV’model,providinguniqueinsightsintotheroleoflanguagedata.Theverticalaxisdepictsanindividualorteammovingbackandforthbetweenthinking,wheretheyreflectondatapreviouslyacquiredanddistillittogainunderstanding,andexperience,wheretheybecomeimmersedinthedataatitssource.Thinkinginvolvesplanningforthegatheringofanynewdatadeemeduseful;andexperienceimpliesmakingthebestuseoflimitedtimetogatherthemostusefulandaccuratedatapossible.

Figure2:

TheRoleofLanguageDatainProblem-solving

AkeypointinFigure2isthat,atthefrontend,languagedataisoftenallthereistoworkwith.DFSSprojectsintheearlystages,forexample,dealwithcustomerandbusinessenvironmentdatathatconsistpredominantlyofstatementsaboutrequirementsandobservationsaboutatargetenvironment.Distillingthatdatahelpsateamfocusontheimportantaspectsoftheopportunityorprobleminordertothinkaboutthenext,moredetailedsetofdatatocollect.Theregionlabeled“II”iswhereDMAICprojectstypicallybeginwithastrongfocusonsomeaspectofapotentiallybroaderissue.

Managing LanguageDataMeasurementVariation

Whilelanguageisdifferentthannumbersinmanyways,therearesomeparallelsinthatdataissometimestaintedwithunwantedvariation.Thisvariationisoftenimposedbythemeasurementsystemusedtocollectthedata.Languageinitsrawform（conversations,email,etc.）isoftencoloredwithemotion,judgment,inferenceandunclearmeasures.Whilethesemayconveysomemeaning,theyoftenrepresentnoiseandmakethemeaningunclear.Toimprovetheaccuracyandrepeatabilityofthemeasurementsystem,semanticistS.I.Hayakawadistinguishesthevalueof“reportlanguage,”whichfocusesonthetraceablefacts（Figure3）.

Problem-solversgenerallyoperatemoreeffectivelywhenqualitativedataisgatheredandprocessedintheformofreportlanguage.Thatdoesn’talwayscomeeasy,aspeopletendtogeneratetheemotion,judgment,etc.reflectedatthetopofFigure3.Ittakesenergyandeffectiveprobingtoacquirethedataneeded.

Figure3:

Removing‘MeasurementNoise’DuringLanguageDataAcquistion

Withoutgivingitaname,peopleroutinelyexercisethepowerfulskillofdistillingcommonaspectsinanumberofspecificeventsorthings,classifyingthemattheappropriatedetaillayerintheirmentaldatabase.Thisiscalledabstraction,whichisacorecommunicationandthinkingskill.Hayakawadevelopedtheusefulnotionofthe“ladderofabstraction”（Figure4）,whichhelpsoneunderstandthatabstractionisn’taboutbeingvague.Itinvolvesprecisegeneralization,theprocessoffindingtherightrungontheladderwithenoughdetailforclarityyetnotsomuchthatthedetailgetsintheway.Peoplewouldfinditdifficulttoconverseorthinkwithouttheseamlessabilitytomoveupanddowntheladder.Noonesays,“I’mgoingtotheapple,banana,grape,strawberrystand,”when“fruitstand”conveystheideawithappropriatelylessdetail.Ontheotherhand,mostpeopledon’tasksomeonetobringthemback“somefood”forlunch.Whiletheseexamplesseemtrivial,theyshowhowautomatictheprocessofabstractionis.Figure4providesaviewofabstractionatthreelevels.Partofthechallengeisfindingtherightlevelforaparticularuse.

Figure4:

TheLadderofAbstraction

Notingthedistinctionbetween“contextdata”and“needsdata”canhelpanyonegatherbetterinformationandenhancethewaythedataisprocessed.Needsdataconveyssomethingvaluabletohaveorbeabletodo.Contextdata,ontheotherhand,referstoobservationsorstatementsaboutanenvironment.Eachofthesekindsofdatacarryimportantinformationthatcanhelpinarticulatingacomplexproblem,developingstatedandlatentrequirements,ordiscoveringfactorsimportanttorobustdesign.Companiesthatunderstandthevaluehiddenincontextdatapayalotofattentiontogatheringandprocessingthatdata.

Focused DiscussionsforBetterLanguageData

Agoodapproachtogatheringlanguagedatathatcontainstherightlevelofabstraction,reportlanguageifappropriate,andtherightmixofusefulcontextandneedsdetailisafocusedopen-endedinterview.AsillustratedinFigure3,movingfromdata“aswefindit”（oftenaninterviewee’sfirstresponse）tothelevelofdetailandclarityneeded,usuallyinvolvesprobingsincemuchoftherichestdataisfoundintheanswerstofollow-upquestions.MoredetailandmanyusefultipsinthisareaareavailableinEdwardF.McQuarrie’sclassictext,CustomerVisits.

Language Data:

The‘OtherData’ofSixSigma,Part2of2:

UsingLanguageData

DavidL.HallowellFebruary26,2010

Part2ofthisarticlefocusesonhelpingpractitionersunderstandthetoolsusedforprocessinglanguagedatainSixSigmawork.Part1 –NatureofLanguageDataisabouttherolelanguagedataplaysinSixSigma,howtounderstandit,includingthetypesandrefinementofthedata,aswellashowtogatheritinthemosteffectiveway.

Anumberofusefulwaysexistforprocessingandusinglanguagedata,amongthempre-processingrawlanguagedataforspecificuses,focusingonanefficientdatasampleanddistillingthedataforparticularpurposesusingtoolslikenet-touch,affinitydiagramsorKJanalysis.

Pre-processing RawLanguageData

Oneprobleminherentinlanguagedataisthevolumeofrawmaterialnecessarytominethemostusefulinformation.Whencapturingvoiceofthecustomer（VOC）dataitishelpfultofirstpre-processinformationtohighlightcontextandneedsdata,asdiscussedinPart1.Figure1illustratesthisforasimplecase.Extractingkeyphrases,whilemaintainingaclearandtraceablelinktotheirsources,isthefirststepinpreparingthedataforfurtherdistillation.

Figure1:

Pre-ProcessingTranscriptstoHighlightContextandNeedsData

Legend:

∙Needs

∙Context

Thecustomerisawarehousemanager.

IT:

Whatkindsofdatadoesthesystemneedtointeractwith?

Customer:

WeneedtobeinconstanttouchwithourMRPsystemandtherelatedinventoryfiles.Ordersincomingfromsalesarereadfromsystemsineitherofthreeserversaroundtheworld.

IT:

Areorderscominginaroundtheclock?

Customer:

Wehope.Keepingupwiththefollow-the-suntimingischallenging.Delaysinourabilitytoconfirmavailabilityanddeliverycreateproblemsforsales…andofcoursewehearaboutthatrightaway.

IT:

Soyourinventory,statusandplanninginformationneedstokeeppacethroughallthreeshifts.

Customer:

That’sright –andifasifthatweren’tenough-thesystemsatourdifferentsalessitesanddesigncentersareatvariouslevelsofcapabilityandstandardization.Weareintothethirdmonthofaglobalupdate –butforanotherthreemonthswewillhavetotalktoamixoftheoldandnewsystemsandtranslatingbackandforthbetweenthem.

IT:

Whenthatstabilizeswillthingsgeteasier?

Customer:

Thatwouldbenice.Butthenitwillbeanewsuppliersystemorsomeotherdatabaseweneedtowriteanewinterfacefor.

IT:

HowdoyouplantheAGVroutesandschedules?

Customer:

Firstthesystemhastogatherupallthedataabouttheproductionplans,theavailabilityandlocationsoftheparts,andthelocation,capaci