Data. Visualization process. Segmented, adapted, chosen data. Sampled or simulated (original) data. Display Data (geometry information)



Similar documents
Data Visualization - A Very Rough Guide

Advanced Diploma of Professional Game Development - Game Art and Animation (10343NAT)

The Evolution of Computer Graphics. SVP, Content & Technology, NVIDIA

The Visualization Pipeline

Volume visualization I Elvins

3D GIS: It s a Brave New World

How To Model Surfaces With Rp

Interactive Level-Set Deformation On the GPU

Creation of an Unlimited Database of Virtual Bone. Validation and Exploitation for Orthopedic Devices

TEXT ANALYTICS INTEGRATION

Unit & Live Testing for SSIS

Architectures for massive data management

Visualisatie BMT. Introduction, visualization, visualization pipeline. Arjan Kok Huub van de Wetering

Interactive Level-Set Segmentation on the GPU

Getting Started with Tableau Server 6.1

Publishing Stories to Lumira Cloud

Best Practises for LabVIEW FPGA Design Flow. uk.ni.com ireland.ni.com

EDI Overview 3. EDIConnect Benefits 3. EDIConnect - A Complete Solution 4. Key Technologies 5. Translator 5. Transaction Builder 7

Computer Graphics AACHEN AACHEN AACHEN AACHEN. Public Perception of CG. Computer Graphics Research. Methodological Approaches

Forecasting Traffic Growth and Impact with Cisco MATE Design

CAD and Creativity. Contents

2933A: Developing Business Process and Integration Solutions Using Microsoft BizTalk Server 2006

SAP PPM Portfolio Analysis Dashboard

SolidWorks Corporation: CSWP Sample Exam

Firewall Verification and Redundancy Checking are Equivalent

AML710 CAD LECTURE 1. Introduction to CAD. Motivation Principles of CAD / CAM. AML710 Computer Aided Design

How To Get A Computer Engineering Degree

Success Factors Recruitment Management

Performance Management and KPIs

How To Create A Data Visualization

PROGRAM OUTLINE PAGE 1

Setup for Success. We encourage you to visualize the following when thinking about PipelineDeals three core concepts:

Scalable Machine Learning - or what to do with all that Big Data infrastructure

Integrated Modeling for Data Integrity in Product Change Management

PCL - SURFACE RECONSTRUCTION

Armanino LLP Welcomes You To Today s Webinar:

Reduce Waste with Spray Technique Training in painting & coating operations

Solving Common Sales & Marketing Problems Using Marketing Automation Software (MAS) by Ann Stanley, Managing Director

Chapter 12. Development Tools for Microcontroller Applications

Appendix M INFORMATION TECHNOLOGY (IT) YOUTH APPRENTICESHIP

Technical white paper.

Water Flow in. Alex Vlachos, Valve July 28, 2010

VIDEO SCRIPT: Data Management

TRANSFORMING LIFE SCIENCES THROUGH ENTERPRISE ANALYTICS

THE FUTURE OF SOCIAL CRM

NewZapp Managed Service

Automotive EMI System

Motivation: Smartphone Market

ADVANCED COMMUNICATION SERIES SPEAKING TO INFORM. Assignment #1: THE SPEECH TO INFORM

Querying MongoDB without programming using FUNQL

Dong-Joo Kang* Dong-Kyun Kang** Balho H. Kim***

Using Power to Improve C Programming Education

OVERVIEW AVISO INSIGHTS

Master of Science in Computer Science

An Introduction to Qualitative Data Analysis (QDA) with Atlas.ti

Industrial Vision Days 2012 Making Cameras Smarter: FPGA Based Image Pre-processing Unleashed

Sales Pipeline/Funnel Management Template User Guide

Real-time Processing and Visualization of Massive Air-Traffic Data in Digital Landscapes

PROJECT MANAGEMENT INNOVATION BASED ON BUILDING INFORMATION MODELING

Avature CRM. Get Engaged to Talent

Exclusive access to metrics to measure KPIs in real time, and at scale

Visualizing Electromagnetic Fields: The Visualization Toolkit. Michael Selvanayagam

Predictive Analytics

credits Programming with actors Dave B. Parlour Xilinx Research Labs Thomas A. Lenart Lund University Robert Esser

Building Information Modelling (BIM); How it Improves Building Performance. R.P. Kumanayake Lecturer, Department of Civil Engineering

Scientific Visualization with Open Source Tools. HM 2014 Julien Jomier

Facts about Visualization Pipelines, applicable to VisIt and ParaView

APD to Update Marketing Attributes from SAP BI to SAP CRM

4.4 Transforming Circles

What's New in Heavy Civil Construction Software. Chris Richardson HCC-6898

Geometry: Unit 1 Vocabulary TERM DEFINITION GEOMETRIC FIGURE. Cannot be defined by using other figures.

XT B-Rep; Making it Real

Service-Oriented Visualization of Virtual 3D City Models

Computers in Film Making

A Flexible Medical Image Archiving and Reporting System

WESTMORELAND COUNTY PUBLIC SCHOOLS Integrated Instructional Pacing Guide and Checklist Computer Math

Reporting in The Financial Edge David Kilmer

Transcription:

Waltz{AnExploratoryVisualizationtoolforVolumeData, usingmultiformabstractdisplays ComputingLaboratory,UniversityofKent,Canterbury,Kent,England,UK JonathanC.Roberts simplifying(orabstracting)thestructureoftheviewedinformation;andlinkingobjectsandviewstogether(allowing toaidauserinterpretdatacorrectly.theseare:displayingthedataindierentforms(multiformvisualization); Although,visualizationisnowwidelyused,misinterpretationsstilloccur.Therearethreeprimarysolutionsintended ABSTRACT correspondingobjectstobejointlymanipulatedandinterrogated). andrene'visualizationcycle. endofalongvisualizationprocess.ourvisualizationsystem,basedonanadaptedvisualizationmodel,allowsa displaymethodtobeusedthroughoutthevisualizationprocess,inwhichtheuseroperatesa`display(correlate) howeverthatcurrentvisualizationsystemsdonoteectivelyutilisethedisplay,forexample,oftenplacingitatthe Thesewell-knownvisualizationtechniques,provideanemphasistowardsthevisualizationdisplay.Webelieve Keywords:Visualization,Visualizationsystem,Multiform,Manipulation,Abstraction befurtherdisplayed,manipulatedandltered. manipulated;asetof`portionsofinterest'canbeselectedtogenerateaspecializeddataset.thismaysubsequently Thisdisplayintegrationprovidesausefulexplorationenvironment,whereobjectsandviewsmaybedirectly ifthepictureisincorrect:byalgorithmicerrors,accumulatedduringtransformations,orinherentinthealgorithm atacertainresolution?then,theusermaymisinterpretormisunderstandtheunderlyinginformation. Theadage\apictureisworthathousandwords"providesapowerfulincentiveforvisualizingdata.However,what 1.INTRODUCTION scientistwantingtodiscoverfacts,auserinvestigatingandclarifyingthescientistsworkoraresearcherpresenting results,theaimsarethesame:toperceiveandgaininsightoftheunderlyingdatabyinvestigatingrelationships, dependenciesandproportionsintheinformation. Understandingphenomenafromtheunderlyingdataisthemainreasonfordoingvisualization.Whether,itisa information.westate:explorationtechniquesfocusingoninteractionmethodsandthevisualizationdisplayare thevisualrepresentationstheusergainsabetter,andhopefullymoreaccurateunderstandingoftheunderlying usefulfor`correctinformationdissemination'. Scienticinvestigationandexplorationisconsequentlyimportant:as,bymanipulatingandexperimentingwith allowtheusertoquicklydisplayarepresentationofthedata,andthenfurtherrenethevisualization. Ourvisualizationsystemprovidesa`display,rene,display'cycleasaseriesofmini-dataowpipelines,which Inreal-life,anexplorer,wishingtonavigatefromCityAtoTownB,wouldusemapsatdierentscales.First,they representationshowingtheminorroads.thelargescalemapprovidescontextinformationforthedetailedmap. wouldlookatamapshowinganoverviewofthemajorroadconnections;thentheywouldlookatadetailedmap 2.THECONCEPTS understandthedetailedinformation. details.thistwo-levelapproachprovidesstructuredinformationtotheuser,whoisthen,inabetterpositionto Likewise,inscienticexploration,itisbesttounderstandtheinformationasawholeentity,thenlookatspecic allowcomplexanddiverseoperations,butdonotprovidetoolsthatencouragetheusertousethisconcept.moreover, :E-mail:J.C.Roberts@ukc.ac.uk;WWW:http://www.cs.ukc.ac.uk/people/sta/jcr/index.html Fromthisexample,weproposethatvisualizationsystemsneedtoadoptthistwo-levelpolicy,currentsystemsdo Otherauthorinformation:

Filter Map Display Data Segmented, adapted, chosen data Display Data (geometry information) Sampled or simulated (original) data Visualization process Figure1.Thetraditionalvisualizationprocess:Filter,Map,Display toachievethisstyleofexplorationonscienticdata,asklinkermentioned,1imagingtoolsneedtoprovideinteractive mechanisms,allowingtheusertointerrogateandunderstandrelationshipsanddependencieswithintheinformation. Webelievethattherearethreeprimary(andpowerful)visualizationtechniquesthataidtheusertoexplore andgenerateacorrectinterpretationoftheunderlyinginformation.first,displaythedataindierentforms{ Multiformvisualization.Second,displaytheinformationinsimpliedAbstractviews.Third,provideahighdegree ofinteractionanddirectmanipulation. Multiformvisualizationistheprocessofdisplayingthesameinformationinmultipleways.2Thisenablestheuser tounderstandtheinformationfromdierentperspectives,helpingtoovercomeconfusionandmisinterpretation frombelievingtheresultfromasingularvisualizationmethod.thishasbeensuccessfullyusedtodisplay multivariatedata,asexempliedbythexmdvtool,3andwe,likeyageletal,4advocatetechniquesthat utilisedierentrenderingschemestouncoverdierentaspectsinthedata. Abstractviewsrepresentsimplied,augmentedorstructuralrealizationsoftheinformation.5Thistechnique generatesmultiformviews,asabove.however,theseviewsaregeneratedbythesamealgorithmwithdierent simplicationcriteria.forexample,asurfacealgorithmmaybegeneratedatahighresolutionwithmany surface-trianglesoralowresolutionwithfewertriangle-elements,thelatterwouldberenderedfasterand providesasimpliedversionoftheinformation.thissimplicationisusefulfor`fastinteractive'navigation, asusedbythevolvissystem,6forexample. Infacttheabstractionmayoccurateitherthedataortheimagelevel.Atthedatalevel,theinformationcontent maybereduced(simplied)bysamplingorgroupinglikeobjectstogether;forexample,silver7segmentsthe dataintogroupsnamedobjects,thatmaybetrackedastheyevolveandmovethroughoutthedata.8at theimagelevel,thevariableorattributesbeingmappedontothedisplaymaybealtered.forexample,the trianglesmakinguptheisosurfacemaybeexchangedtolinesorpoints;thiswouldallowotherwiseoccluded objectstobeseen.alternatively,thepositionoftheobjectsmaybeaugmentedtofacilitatetheunderstanding ofthedata.forexample,sheelaghetal9usethistechniquetoprovidea`clearlineofsight'toobscuredan concealedsectionsofthedata. Theabstractviewsmayallowobjectsthatarenormallyoccludedtobeseen,orgenerateastructuralviewof thewholedata.theyprovidetheuserwithadditionalinformationaboutthestructureofthedata:usefulfor a`correctdissemination'oftheinformation. DirectManipulationallowsobjectswithinthedisplaytobedirectlyinterrogated,scaledandpositioned.This interrogationisalsopropagatedtorelatedviewsofthesamedata,toprovidejointinteractivedataprobing andobjectmanipulation.thisissimilartothebrushingtechniqueofward,3thatallowselementsinoneview tobebrushedbythemouse,coincidentallyhighlightingcorrespondingelementsinrelatedviews;andsimilar tothemanipulationlinkingtechniqueusedintelecollaboration1orbyothersystems.10{12 3.THEWALTZEXPLORATIONMETHOD TheMultiform,AbstractandManipulationmethodsarewell-knownvisualizationtechniquesandtheiremphasisis towardsthedisplayedimage.userswishingtoexplorethevisualizationneedtoconcentrateonthevisualoutput andnotthevisualizationprocess.however,webelievethatcurrentvisualizationsystemsdonoteectivelyintegrate

Traditional connected dataflow pipeline A Data Filter Map Display Filter Map Display Dataflow connected using the display filter method B Data Filter Map Display Filter Map Display thedisplayintothevisualizationprocess,oftenplacingitattheendofalongsequenceofcommands.thislong visualizationprocessisverypowerful,comprehensiveandextensible,andisoftendescribedbythedataowmodel. Figure2.Thedisplayactingasalter,thatmaybefurtherdisplayedandltered data display filter an additional filter specialization operation included visualizationsystemsarebasedonthismodel.14{18 displayed,figure1.thisworkisoftenaccreditedtohaberandmcnabb13andupsonetal14andmanyexisting Where,thedataisltered,tocreateasubsetofthedata,whichisthenmappedintoarepresentationwhichcanbe Thedataowmodeldescribesthevisualizationprocessasaseriesofthreeprocesses:lter,mapanddisplay. thevisualizationtasks,byprovidinganintelligentuserinterface.forexample,woodetal19mentionthethesisof Stead,andtheVISTAsystem20providesaseriesofrulesthatappropriatelydisplaythevisualization.Suchsystems pretation.onesolutionistoshortenthevisualizationcycle.thismaybeachievedbyautomatingsomeorallof donotemphasiseexplorationtechniques.analternativemethodmayconsequentlybegeneratedbyintegratingthe Webelievethislengthyvisualizationcyclemaycausefrustrationandconfusion,leadingtoinformationmisinter- thenbefurtherlteredanddisplayed. displayintothevisualizationpipeline,byallowingthedisplaytoactastheinformationlter.theltereddatamay informationhasbeenspecialized.thisnewdatasetmaybefurtherlteredanddisplayed.thediagram,figure2b, operation,figure2b.thus,thedatahasbeenrenedthroughthisdisplay-lter,andanewdatasetisformedby makingasubsetfromaselectionoftheinitialdataelements.wenamethisnewdatasetadataspecialization,asthe alsoshowsthattheorderoftheprocesseshasbeenalteredtomap,displayandlter.thisallowssomelterprocess Thislteroperation,namedadisplaylter,21allowstheoutputfromonedisplaytobeinputtoanotherlter tobeomitted,enablingtheselectedobjectstobedisplayedwithoutbeingrened.moreover,thespecializeddata maybeadditionallylteredbythereceivingmini-pipeline.obviously,multipledisplay-lteroperationswillgenerate aseriesofchilddata-specializations,whichmayformadata-hierarchyofsubdatasets,figure3. Filter Map Display Filter Map Display Figure3.Ahierarchyofdataselectionsmaybegenerated Filter Map Display Filter Map Display Filter Map Display

Thedatawithinanyofthedata-specializationsmaybedisplayedinmultipleforms,consequentlygeneratinga groupofrenderers.4anothervisualizationmethodmaybeaddedtothisdisplay-groupbysimplyrequestingthe appropriatemethod. Themanipulationofelementsisalsoscopedbytherender-group.Thisallowsthesameobjects,butindierent views,tobejointlymanipulated.forexample,objectswithinasurfaceandaskeleton-displayofthesameinformation maybejointlymanipulated:scaled,rotatedortranslated.ourmanipulationoperationsjointlytransformtheviewing projection(inthreedimensions)orhighlightandtransformindividualelements.additionally,wegrouptogetherthe mapping-attributes,sothehue,colourortransparency(ofthemappedobjects)maybecoincidentlyaltered. Thisstyleoflinked-manipulationissimilarlyimplementedinsystems,suchas,Klinker1inAVS,KBVision,10 LinkWinds11andSciAn,12but,ratherthanallowtheusertojoinappropriatelinks,ourdisplay-groupmethod allowsthelinkingtobeautomaticallyprovidedbythesystem.thelinksareautomaticallyaddedtotherelevant displayswhenanewdisplayisaddedtothelist,thussimplifyingtheuserinteraction. Theuserinterfaceisfurthersimpliedbycoercingtheprocessesofthemini-visualization-pipelineintoonemodule. Thisonemoduleallows,throughmenuselections,thelter,mappinganddisplayoperationstobeperformed.Thus, tocreateanewmoduletheuserselectstheexportmenu,thatautomaticallygeneratesanothermodule.thistechnique couldbedescribedasbeingamixtureofamodulebuildingenvironmentandamenu-basedvisualizationsystem; becauseoftheuser-controllablenetwork,withmenu-basedcommands. Further,wehavedevelopedamodule-manager,thatautomaticallypositionsthemodulesontheworkarea,taking theresponsibilityfromtheuser.anewwindowisgeneratedforeachdisplay,sothemanagerlabelseachwindow withauniquereferencenumber,relatingtheseparatewindowstothework-area.themultiformdisplayspersist throughoutavisualizationsession,enablingthedisplaystoprovidea`history'ofpreviousexperimentations. Oursystemincorporatesthedisplayasarst-class-citizen,byincludingthedisplayoperationasanintegralpart ofthevisualizationprocess.thisencouragesandsupportstheuserto,rst,displaythesamedata,andaspectsof thedata,usingdierentrenderingschemes.second,providedirectmanipulationoftheviewsandobjectswithin resultantvisualizations,toallowtheusertodirectlyselectobjectswithinthedatatoenablethedatatobeltered. Third,toautomaticallylinkcorrespondingfeaturesbetweenvisualizations,allowingobjectsandviewstobejointly manipulated.thevisualizationprocessthenbecomes:aprocessof`display(correlate)andrene'operations. 4.EXAMPLES{AUSER'SVIEW TheWaltzsystemhasbeendevelopedforvolumedata.Webelievethemethodologycouldbeeasilyextendedto otherdata-types,alongwiththehigh-leveluser-interfaceimplementation.however,thelow-leveldatastructure wouldneedtobegeneralizedtosupportotherdatatypes. Todescribetheoperationandcontrolofthesystem,wewillworkthroughtwoexamplesessions.Therstexample usesatestdatasetcontainingconcentricthree-dimensionallayers. Onstartup,acleancanvasispresentedtotheuser.Arstmoduleisautomaticallyplacedonthecanvaswhen thedataisloaded.thisrstmoduledisplaysthele-name,holdstheimporteddataandactsastherootnodeon theworkarea.subsequent`children'modulesarealwaysofthedisplay-grouptype.thelteroperationontherst display-groupmoduleisusedtoprocessthedataelementsintoobjects.currently,onlyonealgorithmisprovided, thatclustersthe(individual)dataelementsintosetsofsimilarvalue.thealgorithmusesaoodingtechniqueto thresholdandconnectthedata.initially,noimagesaredisplayed,butadditionalrealizationsareadded(anddeleted) byselectingtheappropriatedisplay-groupmenu.initiallyweaddasurfacerendering;however,asthewholedataset isanobjectitself,athree-dimensionalcubeisrendered,andtheinternalstructureremainsunseen.therefore,we addaslice-display,enablingtheinternalstructuretobeviewed,figure4. Throughourexplorationwerealizethattherearemultiplepartstothedataset;wecaneitherre-evaluatethedatasegmentationinthismoduleoraddanothermodule:asachildofthedata-lemodule.Infact,whenthesegmentation isre-evaluated,eachdisplaymethod(withinadisplay-group)automaticallydisplaysthenewinformation. Weaddanothermodule,Figure5,thistimeweprocessthedataintotwogroups:theinteresting-featureand thesurroundingair.thefeatureisselected,consequentlyhighlightedbythesystem,andafurtherdisplay-module isgeneratedholdingonlythisfeature,towhichweaddasurface-displayandalist-display,figure6.thelistmoduledisplaysthreegroups:theobjectandtwopaddinggroups(alltheelementsoutsidethisandtheprevious specialization).

displaywindow) Figure4.TheWaltzworkarea,showingthedata-lemodule,agrouped-displaymodule(withasliceandsurface- Figure5.Twonewgrouped-displaymodules,evaluatedintotwointeresting-feature,shownbytheslice-display intoafourthgroup-displaymodule,andasurface-displayisaddedtothismodule.bymethodssimilartothisthe aslice-display,figure7.thensomeoftheelements,specically3,4,5,6,7asseenfromthelist-display,areexported usercanexploreandfocusonspecicpartsoftheinformation. Finally,were-evaluatethegrouped-featuresonthisthird-moduleandincludeabounding-boxabstractionand wishtobothdisplaythemainsurfaceandinvestigatethepressureswithinthedata. isbombardedwithasmallerpieceofmaterial.thedatasetdescribesthepressureintheblock.fromthisdata,we Initially,wedividethedatasetintotwointerestingfeatures:theblockandair,Figure8.Boththesliceand Thesecondsessionvisualizesthedatafromaspacedustimpactsimulation;whereastationaryblockofmaterial grouped-displaymodule.thissecondmoduleisusedtodisplaytheblocksurface. surface-displayareusedtorealizethefeatures.theblockisthenselectedandexportedtogeneratedthenext objectspecializationinthethirdmoduletogeneratemultipleobjects.towhich,weaddsurface,bounding-boxand network-displays,seefigure9. Insimilaroperationtogeneratingthesecondmodule,wecreateathirdmodule.Were-segmentthesingle- Userinterrogationmaybeperformedonthedata.Forexample,theobjectsmayberotated,Figure10or

Figure6.TheWaltzworkarea,showingthedata-lemodule,agrouped-displaymodule(withtwoabstractions: theslice-displayandasurface-display) transformedfigure11.theinteractionisautomaticallydirectedtotherelevantobjectsofeachrendererina Bounding-box,Surface,listandslicedisplays. Figure7.Thedatainthethirdmoduleisrenedfromtheone-groupintosixgroups,andrealizedusingthe particulardisplay-group. ThenameWaltziscoinedfromthemusictermofthreepartsinfour,fromtheMultiform,AbstractandManipulation processeswithinagraphicaluser-interface. WaltzhasbeendesignedinC++withclassesrepresentingthedata,groupingsofthedataandabstractions, 5.IMPLEMENTATION andisimplementedoverxandxlibusingthemotifwidgetset,inventor22andopengl23libraries.foradetailed descriptionoftheimplementation,see.5

exported.thesecondmoduledisplaysthesurfaceoftheblock. Figure8.Therstmoduleisusedtosegmentthedataintotwofeatures;theimpact-blockisthenselectedand Sowhenanewmoduleiscreatedthemanagerpositionsthewidgetsinthedesiredposition.Likewise,asnew Themodule-managercontrolsthepositionsofthemodulesandiscontrolledusingtheXcallbackmechanism. Figure9.TheWaltzworkareashowingonedata-lemodulewiththreegroup-displaymodules. display-abstractionsareaddedtothedisplay-groupthemodule-manageriscalledtoupdateworkarea. hascontrol-structuresallowingupanddowntree-traversal. ulators.threedimensionalmanipulators(jacks)areappliedtotheobjectsintheinventorscene-graph.togenerate ThemanipulationwithinthedisplaysiscontrolledusingboththeXcallbackmechanismandtheInventormanip- Themodule-managerstoresthemodulehierarchyinformationinatreedata-structure.Forconveniencethetree thelinkedmanipulationwithinthegrouped-displaymodule,thecodereferencesoneinventor-manipulator,thatis placedineachdisplay-method'sscenegraph.infact,individualmanipulatorscanbeswitchedintothegraph,to allowspecicmanipulation.thisswitchiscontrolledbyaradio-buttonselectiononamotif-basedform. to24),withalistofthegroupsandtheiradjacencyinformation. data.relevantpartsofthedataarecopiedandexportedfromonemoduletothenext.thisexportoperationis controlledexplicitlybythesystemuser.eachdatashadowholdsthegroupingsdataasanarrayofslices(similar ThedataisstoredinaDataShadowclass,thatholdsinformationaboutthegroupingsandspecializationsofthe

Figure10.Thewholeviewmaybedirectlymanipulated.Thecontrolislinkedbetweenviewswithin thesamedisplay-group.theimagesshowthreeviewsbeingrotated. Figure11.Individualdataelementsmaybedirectlytransformed.Thecontrolislinkedbetween viewswithinthesamedisplay-group.theimagesdepictonegroupbeingtranslated.

Wehavepresentedasystemthatprovidesexplorationtechniques,especiallyforthe`correctdissemination'ofinformation.Weacknowledgethatexplorationisonlyonemethodtosuppressmisinterpretation.Butwebelieveittobe 6.DISCUSSIONANDFUTURERESEARCH animportantone!furtherworkisrequiredinthis`correctdissemination'area. ItistruethattheModulevisualizationenvironments(AVS,IRISExploreretc,25)allowmultipleviews,butinour experiencethesetendtobeexpensive:usingtime,processingandmemoryresources.thus,therearesomeadvantages inhavingmultiplelightweightrenderingwindows,suchasprovidedbywaltz. Oursystemprovidesmethodsthatencouragestheusertoperformmultipledisplaysofthesameinformation. ofpreviousexperimentations.infactwebelievethatabstracttechniquesareuseful,becausetheyprovidesimpler anddiverseinterpretationsoftheunderlyinginformation.althoughsomeabstracttechniquesaremeaninglessin occludedelements,forexample,tobeselectedandmanipulateddirectly. isolation(suchasthelist-display)theycanactascontrol-abstractions,andwhenlinkedwithotherviewstheyallow Themultiformdisplayspersistthroughoutavisualizationsession,enablingthedisplaystoprovidea`history' features,andtomanagethecontrolofthelinked-interactionandtheplacementofthemodules.othersystems,such asthetelecollaborativeenvironmentdescribedbyklinker1orthevisualizationinputpipeline26provideextensible webelievethatitisusefultoallowtheusertolterthedatathroughthedisplay,bysegmentingtheinformationinto interactivevisualizationsystemsbutdonotmanageeitherthelinked-interactionorthemodule-layout. OthersystemssuchasSciAn12andLinkWinds11providemultipleviewsofthedatainalinkedenvironment,but, neatlayout!problemswiththislayoutwouldoccurinalongsession.currently,modulesmaybedeletedandthe together,andreducestheirsize,toallowlargernetworkstobevisualizedonthework-area. workareascrolled.butitwouldbebenecialtoapplyamorecomplicatedalgorithm,that(say)groupsthemodules byreducingtheneedforadditionalpipesorlinksandstopstheuserfrom`wastingtime'movingthemodulesintoa Webelievethat,bymanagingthemodule-layout,theuserreceivesadditionalinformation,simpliesthenetwork example,thevistasystem20,27operatesbygeneratinganappropriaterealizationfromalistofordered,anduser preferencedvariables.thistechnique,maybeusefulfornaiveandinfrequentusersbutexpertusersmaywishfor morecontroloverthegeneratedvisualization.indeed,misinterpretationsgeneratedthroughthewrong-useofthe visualizationprocesswouldbesignicantlyreducedusinganintelligentsystemapproach. Theintelligentvisualizationsystemstakethesystemandinformation-managementtechniquetotheextreme.For ltertechnique.thedisplay-ltermethodprovidesanintuitivelteringmethod,albeithardertoimplementand technique,withsixdisplaymethods.othertechniqueshavebeenproposed,butareleftforfuturework.moreover, lessdiversethantraditionaltechniques.infact,thecurrentimplementationonlyprovidesonedata-segmentation theabstractiontechniquecouldbeappliedmorevigorouslytothecurrent(andfuture)displaymethods.thecode, However,anintelligentapproachdoesnotallowsomuchuser-exploration;suchasprovidedthroughadisplaytechniqueanewsub-classisrequired,andthesystemre-compiled. wasdesignedtobeextensibleespeciallytheuserinterface,butinpractice,foreachnewsegmentationordisplay informationinterpretation. displayedrealizations.comparativevisualizationtechniques28mayprovideadditionalinformation,aidingacorrect Themultiformlinkeddisplaysprovideausefulexplorationenvironment,however,theusermanuallycomparesthe fromitsoriginalpositionandautomaticallyreplaceditwhentheuserreleasedcontrol. intocompoundoperations,forexample,anoperationmaybedesignedto`lookandsee',thatmovedafeatureaway niquescouldbeimplemented.currentresearchistakingplaceinthisarea.someofthemechanismscouldbemade Waltzcurrentlyimplementsselectionandtransformationstylesoflinking.Furtherbrushing3andlinking1tech- WehavepresentedavisualizationsystemincorporatingMultiform,AbstractandManipulationtechniques,forthe correctdisseminationofscienticinformation.wehaveadvocatedtheuseofexplorationtechniques,andacknowledge thatexplorationisonlyonemethodtosuppressthismisinterpretation. 7.CONCLUSION visualizationandcorrectdisseminationinauserextensibleframework. WeproposetodevelopandextendtheWaltzsystem,toincorporateandprovidemechanismsforexploratory

8.ACKNOWLEDGEMENTS IacknowledgeSteveHillforhisadviceonthisproject,TimHopkinsforhishelp,DuncanLangfordforproofreading thispaperandtheuniversityofkentphysicslaboratory,u.k.fortheimpactdata. REFERENCES 1.G.J.Klinker,\Anenvironmentfortelecollaborativedataexploration,"inProceedingsVisualization'93{ sponsoredbytheieeecomputersociety,pp.110{117,1993. 2.P.R.KellerandM.M.Keller,VisualCues{PracticalDataandVisualization,IEEEComputerSocietyPress, 1993. 3.M.O.Ward,\XmdvTool:Integratingmultiplemethodsforvisualizingmultivariatedata,"inProceedingsVisualization'94{sponsoredbytheIEEEComputerSociety,pp.326{333,1994. 4.R.Yagel,D.S.Ebert,J.N.Scott,andY.Kurzion,\Groupingvolumerenderersforenhancedvisualizationin computationaluiddynamics,"ieeetransactionsonvisualizationandcomputergraphics1,pp.117{132, June1995. 5.J.C.Roberts,AspectsofAbstractioninScienticVisualization.Ph.Dthesis,KentUniversity,Computing Laboratory,Canterbury,Kent,England,UK,CT27NF,October1995. 6.R.S.Avila,L.M.Sobierajski,andA.E.Kaufman,\Towardsacomprehensivevolumevisualizationsystem,"in ProceedingsVisualization'92{sponsoredbytheIEEEComputerSociety,pp.13{20,1992. 7.D.Silver,\Object-orientedvisualization,"IEEEComputerGraphicsandApplications15,pp.54{62,May1995. 8.R.Samtaney,D.Silver,N.Zabusky,andJ.Cao,\Visualizingfeaturesandtrackingtheirevolution,"IEEE Computer27,pp.20{27,July1994. 9.M.Sheelagh,R.Carpendale,D.J.Cowperthwaite,andF.D.Fracchia,\Distortionviewingtechniquesfor3- dimensionaldata,"inproceedings'96informationvisualization{sponsoredbytheieeecomputersociety, pp.46{53,1996. 10.AmerinexA.I.,Inc,409MainStreet,Amherst,MA01002,GeneralSupportToolsforImageUnderstanding, 1992.(http://www.aai.com/AAI/KBV/KBV.html). 11.A.S.Jacobson,A.L.Berkin,andM.N.Orton,\LinkWinds:Interactivescienticdataanalysisandvisualization,"CommunicationsoftheACM37,pp.43{52,April1994. 12.E.PepkeandJ.Lyons,SciAn:User'sManual.SupercomputerComputationsResearchInstitute,FloridaState University,Tallahassee,Florida,1993.(http://www.scri.fsu.edu/lyons/scian). 13.R.B.HaberandD.A.McNabb,\Visualizationidioms:Aconceptualmodelforscienticvisualizationsystems," invisualizationinscienticcomputing,b.shriver,g.m.nielson,andl.j.rosenblum,eds.,pp.74{93,ieee ComputerSocietyPress,1990. 14.C.Upson,T.Faulhaber,D.Kamins,D.Schlegel,D.Laidlaw,F.Vroom,R.Gurwitz,andA.vanDam,\The applicationvisualizationsystem:acomputationalenvironmentforscienticvisualization,"ieeecomputer GraphicsandApplications9(4),pp.30{42,1989. 15.G.AbramandL.Treinish,\Anextendeddata-owarchitecturefordataanalysisandvisualization,"inProceedingsVisualization'95{sponsoredbytheIEEEComputerSociety,pp.263{270,1995. 16.D.S.Dyer,\Adataowtoolkitforvisualization,"IEEEComputerGraphicsandApplications10(4),pp.60{69, 1990. 17.SiliconGraphicsComputerSystems{SiliconGraphicsInc.,IRISExplorerTechnicalOverview,April1992. (http://www.nag.co.uk/visual/ie/iecbb/docs/technicalreport.ps). 18.M.Young,D.Argiro,andS.Kubica,\Cantata:VisualprogrammingenvironmentfortheKhorossystem," ComputerGraphics29,pp.22{24,May1995. 19.J.Wood,K.Brodlie,andH.Wright,\Visualisationovertheworldwidewebanditsapplicationtoenvironmental data,"inproceedingsvisualization'96{sponsoredbytheieeecomputersociety,1996. 20.H.SenayandE.Ignatius,\Aknowledge-basedsystemforvisualizationdesign,"IEEEComputerGraphicsand Applications14,pp.36{47,November1994. 21.W.J.Schroeder,W.E.Lorensen,G.D.Montanaro,andC.R.Volpe,\VISAGE:anobject-orientedscientic visualizationsystem,"inproceedingsvisualization'92{sponsoredbytheieeecomputersociety,pp.219{226, 1992.

22.J.Wernecke,TheInventorMentor:ProgrammingObject-Oriented3DGraphicswithOpenInventor,Release2, 24.K.J.ZuiderveldandM.A.Viergever,\Multi-modalvolumevisualizationusingobject-orientedmethods,"in 23.J.Neider,T.Davis,andM.Woo,OpenGLProgrammingGuide{TheOcialGuidetoLearningOpenGL, Release1,Addison-Wesley,1994. 1994SymposiumonVolumeVisualization{Washington,D.C.,October17{18,A.KaufmanandW.Krueger, 26.W.FelgerandF.Schroder,\Thevisualizationinputpipeline{enablingsemanticinteractioninscienticvisualization,"inEurographics'92(ComputerGraphicsForumVolume11No.3){AlistairKilgourandLars 25.C.Williams,J.Rasure,andC.Hansen,\Thestateoftheartofvisuallanguagesforvisualization,"inProceedings Visualization'92{sponsoredbytheIEEEComputerSociety,pp.202{209,1992. eds.,pp.59{74,acmsiggraphandieeecomputersocietypress,1994.(vroom). 28.H.-G.PagendarmandF.H.Post,\Comparativevisualization-approachesandexamples,"inFifthEurographics 27.J.Mackinlay,\Automatingthedesignofgraphicalpresentationsofrelationalinformation,"ACMtransactions ongraphics5(2),pp.110{141,1986. KjelldahlEds.,pp.139{151,BlackwellPublishers,1992. WorkshoponVisualizationinScienticComputing,May1994.