【文档说明】Information Visualization chapter 课件11.ppt,共(28)页,1.427 MB,由小橙橙上传
转载请保留链接:https://www.ichengzhen.cn/view-44908.html
以下为本文档部分文字说明:
KoreaUniversityDivisionofinformationmanagementengineeringUserInterfaceLabInformationVisualizationchapter11Thinkingwit
hVisualizationKoreaUniversityDivisionofinformationmanagementengineeringUserInterfaceLabContents•MemorySyst
ems•EyeMovements•ProblemSolvingwithVisualizations•CreativeProblemSolvingKoreaUniversityDivisionofinformationmanagementengineeringUserInterfaceLa
bIntroduction•인간이정보를찾는데있어서노력과시간및다른활동들과의상호작용에의한기회비용을최소화하는것은동물이음식물을찾기위한행동과비슷함(경제학적논리)•인간과computer사이에있어서시각화는internalinte
rface로간주될수있고,문제해결전략에는computer-aideddesignprogram이도움이됨•Twowaysofvisualizationssupportthinking-visualqueriesoninform
ationgraphics-extendingMemory•Keypointofthischapter-thetheoryofhowwethinkwithvisualizations·memoryandattentionsu
bsystem·visualthinkingasasetofembeddedprocess☞guidelinesareprovidedfordesigningvisualdecisionsupportsystemsKor
eaUniversityDivisionofinformationmanagementengineeringUserInterfaceLabMemorySystems•Threetypesofmemory-iconi
c,long-term,workingmemory•Visualworkingmemory-holdsthevisualobjectsofimmediateattention(comuter중앙처리장치내의고속기억부)·capacityisperhaps3to5simpleob
ject·positionsofobjectsarestoredinanegocentricmap&ninelocationsarestored※egocentric:viewdorperceivedfromone’sownmindasacenter·th
etimetochangeattentionisabout100msec·semanticmeaningorgistofanobjectorscenecanbeactivatedinabout100msec·foritemstobeprocessedintoLTM,
deepersemanticcodingisneedeed-workingmemoryhasanumberofinterlinkedbutseparatecomponents·auditory&visual,subsystemsforbodymovements
(verbaloutput,motorcontroletc)·BaddeleyandHitch’smodel:controlstructure(centralexecutive)·ACT-R&EPIC(executiveprocessin
teractivecotrol)model:overviewmodelsofcognitiveprocesses→nextpageKoreaUniversityDivisionofinformationmanagementen
gineeringUserInterfaceLabMemorySystems(continue)KoreaUniversityDivisionofinformationmanagementengineeringUserInterfaceLabMemorySystems(cont
inue)•VisualworkingmemoryCapacitya.0.4초간시연후에0.5초의blanktime적용을하고똑같은pattern을보여주되어느한object의색깔이나모양을변경시킴:대부분의결과는3개의object가error없이기억됨b.subj
ectnobetterthanifthestemandthecaparecombinedthaniftheyareseperatedc.combinedwithconcentricsquares,six
colorscouldbeheldinvisualworkingmemory,butiftheywereputinsidebysidesquares,onlythreecolorscouldberetained※iflonger
viewingwaspermitted:uptofiveobjectsafterafour-secondpresentationKoreaUniversityDivisionofinformationmanagementen
gineeringUserInterfaceLabMemorySystems(continue):ThetheoryofvisualworkingmemoryandtheresultsofVoreletal.(2001)s
uggestthatthreeoftheformerglyphscouldbeheldinvisualworkingmemory,butonlyoneofthelatter-ObjectFile,CoherenceFields,a
ndGist·objectfile:temporarygroupingofacollectionofvisualfeaturestogetherwithotherlinkstoverbal-propo
sitionalinformation·gist:describeakindofsummaryofthepropertiesofanobjectorascene,consistsofbothvisualinformationaboutthetypicalstructu
reofanobjectandlinkstorelevantverbalpropositionalinformationKoreaUniversityDivisionofinformationmanagementengineeringUserInterfaceLabMemorySyste
ms(continue)-ChangeBlindness:visualworkingmemory의제한된capacity때문에하나의장면에서다음장면사이에아주큰변화를주는것이가능하고,그변화가최근에주의를준것이아니면사람들은보통그것을인지하지못한다(unlessthechangeistosome
thingtheyhaverecentlyattended)-Spatialinformation·9개의위치가기억가능하나그것들중에3개가objectfile과연결되어있으며,나머지는그위치에무언가가있었다는것만으로인식됨·Movingtarget의경우4~5개의물체에대하여지속
적인추적이가능함KoreaUniversityDivisionofinformationmanagementengineeringUserInterfaceLabMemorySystems(continue)-Attention·inat
tentionalblindness:십자모양의길이에주의를기울일것을사전에알린후실험시피실험자들이거의완벽하게변화를감지하지만,주의를주라는사전경고없이실험을하였을경우에는변화를감지하지못함,단피실험자들의attention의출발이위의예상치못한변화라면감지
가가능함·pullcueandpushcue:pullcue는display상에새로운object가나타날경우그것에주의를기울이게되고(pullsattentiontowardit),pushcue는어디에새로운것이나타날것이라는것을화살표와같은심볼로서시현시켜주는것임→pullcue의
경우약0.1,pushcue의경우0.2~0.4초가소요됨·선택적인주의선택및Stroopeffect→nextpageKoreaUniversityDivisionofinformationmanagementengineeringUserInterfa
ceLabMemorySystems(continue)KoreaUniversityDivisionofinformationmanagementengineeringUserInterfaceLabMemorySystems(continue)-TheRoleofMotioninAttrac
tingAttention·eyemovement가진행되는중에는peripheralvision(주변시)에의한변화감지는miss되기때문에그에대한solution으로서두가지를제시→useauditorycues,blinkingormovingicons•Rensink’sMod
elKoreaUniversityDivisionofinformationmanagementengineeringUserInterfaceLabEyeMovements•Threeimportanttypesofeyemovements-saccadic,smooth-pursui
t,convergentmovements-saccadicsuppression•Accommodation(조절)-관찰자와target간의거리가변함에따라눈은refocus또는accommodate되는데,accommodation반응은약0.2초가걸린다.•EyeMovements,
Search,andMonitoringKoreaUniversityDivisionofinformationmanagementengineeringUserInterfaceLabEyeMovements-supervisorycontr
ol·복잡하고인간에의해간접적으로조정되는반자동시스템을위한용어(ex.비행기에서의warningsystem,displaysforpowergrid)·인간과computer간의interface가supervisorycontrolsyst
em화함에따라visualattention은매우중요함-visualmonitoringstrategies·visualscanning전략은다음과같은요소에의해결정됨→channels,event
s,expectedcost•Long-TermMemory-HumanLTMcanbecharacterizedasanetworkoflinkedconcepsKoreaUniversityDivisionofinformationma
nagementengineeringUserInterfaceLabEyeMovements-TheforgetingofinformationfromLTMisthoughttobemoreofalossofaccessthananerasureofthememorytrace-Chunks
andConcepts·chunking:theprocessofgroupingsimpleconceptsintomorecomplexones·concept:가설검증과정에서형성→newconceptsar
edistinguishedfromtheprototypebymeansoftransformation·whataboutpurelyvisualworkingmemory?→verballongtermmemory를특징짓는추상적인개념들의ne
twork는아니다.·methodofloci:well-knownmnemonictechnique·datamountain처럼category별로조직화하는것이도움이되며,새로운정보를배울때는반복해서노출시키는것보다는기존재하는정보와통합시키는것이효과적
이다·특히verbal과visualconcepts을함께조합하는것이효과적이며,이것은교육에있어서multimedia사용하는중심적인전제이다.KoreaUniversityDivisionofinformati
onmanagementengineeringUserInterfaceLabProblemSolvingwithVisualization-Figure11.11providesanoverviewofthevariouscomponents·이그림에서중간단
계는object와pattern을인식하는데있어서기초를제공한다·제일상위의단계에서visualquery를수행하기위한중요한pattern과object가분석됨KoreaUniversityDivisionofinformat
ionmanagementengineeringUserInterfaceLab•VisualProblemSolvingProcessProblemSolvingwithVisualization1.Theproblemsolvingstra
tegy2.Visualqueryconstruction3.Thepatternfindingloop4.Theeyemovementcontrolloop5.Theintrasaccadicscann
ingloopKoreaUniversityDivisionofinformationmanagementengineeringUserInterfaceLab•ImplicationsforInteractiveVisualizat
ionDesignProblemSolvingwithVisualization-datadisplaysystem에서가장중요한것들·시각처리중간단계(middlestage)에서pattern을찾는능력에advantage를주어야함·problem에대한thinking이
interface에대한것이되면안됨·interface는저렴한비용으로빠른정보검색이가능하여야한다-VisualQueryPattern·visualquery가error없이빠르게수행되기위해서는visualworkingmemory에저장가능한simple한o
bject나pattern이어야한다·3개의최소단위나복잡한1개이상의query가visualworkingmemory에저장가능하며,용량을초과할경우에는다른인식전략이필요하다KoreaUniversity
DivisionofinformationmanagementengineeringUserInterfaceLab•CostsofNavigationProblemSolvingwithVisualization-가장이상적인vi
sualization은singlehigh-resolutionscreen에모든정보가이용가능토록하는것임,만약정보가각각의다른window에표시가되더라도한번의eyemovement만이필요하며조금더나쁜경우는머리의움직임이추가될것이다.-HypertextLink·h
ypertextlink를클릭하는데는1~2초의손의움직임과마우스클릭이관련되고완전히새로운정보화면을가능케한다.따라서인식이reorientation되는데몇초가소요된다-Brushing,DynamicQueries,andHoverQueries·brushi
ng과dynamicqueries는둘다연속적인mouse움직임이어떤data에대한정보를알려줌·hoverqueries는dataobject에mouse를drag하면추가적인정보를popup시켜알려줌·위세가지모두는효과적이기는하나dataspace에서눈의움직임
과같이각각의point사이를jump할수는없다KoreaUniversityDivisionofinformationmanagementengineeringUserInterfaceLabProblemSolvingwithVisualization-Walk
ingorFlyinginVirtualReality·가상의정보공간에서walking또는flying을통한navigation은시간적으로느리고인식적인요구사항이더많음·Table11.1은각각의navigationte
chnique과관련한대략적인시간과cognitivecost를제공함·느린navigationtechnique에서시간대비cognitivecost가높은이유는visualandnonvisualworkin
gmemory에서의내용의손실이나붕괴때문이다.KoreaUniversityDivisionofinformationmanagementengineeringUserInterfaceLab•Magnif
yingWindowvs.ZoomingProblemSolvingwithVisualization-Zoom이용시·zoomin&lookonepattern·holdthepattern(VWM)·zoomout&compareinVWM·matchisfound
thenzoomback·cofirmdetailsofthematch-두개의다른window·eyemovements를통해빠른match가가능KoreaUniversityDivisionofinformationmanagementengineeringUserInterface
Lab-visualworkingmemorycapacity는비교를위한필요한방문을결정하는중요한요소임-만약targetpattern이visualworkingmemorycapacity에저장가능한단순한pattern이면zooming이
더효과적일것이고,targetpattern이3개이상의item일경우는반복적인zoomin&zoomback이필요하기때문에multiwindow가더빠를것이다ProblemSolvingwithVisualizationKoreaUniversityDivisionofinformation
managementengineeringUserInterfaceLab•InterfacestoKnowledgeStructuresProblemSolvingwithVisualization-비
록visualicon이나object가verbalinformation을활성화시키는데빠르고효과적이지만그것은위의정보들간의image와의미가강하고사전에배운것일때효과적인것임-ConceptMapsandMindMaps·지식구
조(knowledgestructure)를나타내는복잡한방법으로node-linkdiagram이사용됨→학생들의학습을돕는conceptmapormindmap으로사용,또는어떤group의공통된이해를구축하기위해서도사용됨-Constellation(
배열)·높은상관관계에있는시각화된node-link는아주복잡한semanticnetwork에대한view제공·constellationsystem은hoverqueries,커서움직임에따른nodehighligting,zoom
등의기법을사용하여많은양의semanticinformation이아주빨리처리될수있음·degree-of-interesttree→많은양의data가tree구조에삽입될경우이해하거나다루기가어렵게되고전체적인통찰력을잃게됨,따라서통찰력을잃지않으면서자세한정보를볼수있는사용자의관심의정도에따른
다양한detaillevel을가진representingmethod를적용KoreaUniversityDivisionofinformationmanagementengineeringUserInterfaceLabProbl
emSolvingwithVisualizationKoreaUniversityDivisionofinformationmanagementengineeringUserInterfaceLabProblemSolvingwithVisualizationKore
aUniversityDivisionofinformationmanagementengineeringUserInterfaceLabProblemSolvingwithVisualizationKoreaUniversityDivis
ionofinformationmanagementengineeringUserInterfaceLabProblemSolvingwithVisualization-LinkingComputer-BasedAnalysiswithVisualization·
multimentionalscaling,SPIREsystem,ThemeScapeKoreaUniversityDivisionofinformationmanagementengineeringUserInterfaceLabProblem
SolvingwithVisualization-TrajectoryMapping·multidimentionalscaling이concept들간의연결(link)을보여주지못하는데반해trajectorymapping은분명한concept들간의관
계를보여준다·실험은concept들간의cognitivepathway를생산토록design되었으며,strongpathway는weak한것과구별될수있다.KoreaUniversityDivisionofinformationmanagementeng
ineeringUserInterfaceLabCreativeProblemSolving-creativethinking은세가지과정으로나눠지고visualization은세단계에모두도움이된다→preparation,production,judgment-
preparationstage·acquiresthebackgroundinformationneededtobuildasolution·visualqueries는중요한pattern을찾는데있어서looselydefined되며,진행됨에따라더욱focus가이
루어짐-productionstage·theproblemsolvergeneratesasetofpotentialsolutions·studyofarchitecturalperspectivedrawings→precisedrawings,realisticallyshadedi
mage,sketch-judgementstage·theproblemsolveranalyzesthepotentialsolutions·제시된많은가설들은pattern을분석하는과정에서도태됨·onlyaf
ewwillbenovelandevenfewerwillleadtoapracticalsolution