WIXLIB

contributed articlesidentifying these principles is basedon three main objectives:Style independence. In order toidentify a general set of principles wecould apply to a variety of complex3D objects, we looked for visual techniques common across different artistic styles and types of objects;Generative rules. To ensure that wecould apply the principles in a generative manner to create cutaways orexploded views, we formed explicit,well-defined rules describing whenand how each principle should be applied. We designed the rules to be asgeneral as possible while remainingconsistent with the evidence from theexample illustrations; andPerceptual/cognitive rationale. Wemotivated each principle by hypoth-esizing a perceptual or cognitive rationale explaining how the conventionhelps viewers better understand thestructure of the 3D object depicted.Through this analysis, we identified a set of general, perceptuallymotivated design principles for creating cutaways and exploded views. Forinstance, the size and shape of cutsin a cutaway illustration are often designed to not only reveal internal partsbut to help viewers mentally reconstruct any occluding geometry thathas been removed. Thus, illustratorscut radially symmetric objects withwedge-shape cutaways that emphasize the object’s cylindrical structure.Similarly, rectangular objects are cutwith object-aligned cutting planes,or box cuts; skin-like covering sur-Figure 6. A general-purpose computer-generated map of San Francisco (left) is notan effective destination map because it is cluttered with extraneous information andneighborhood roads disappear. Our destination map (right) includes only the relevanthighways, arterials, and residential roads required to reach a destination. The layout andrendering style further emphasize the information required to reach it.64communi cations o f th e acm April 2 0 1 1 vol . 5 4 no. 4faces are cut using window cuts; andlong tubular structures are cut usingtransverse tube cuts. Illustrations ofcomplex mathematical surfaces often use exploded views in which eachslice is positioned to reveal local geometric features (such as symmetries,self-intersections, and critical points).We have also examined “how thingswork” illustrations designed to showthe movement and interaction ofparts within a mechanical assembly.The hand-designed illustrations oftenuse diagrammatic motion arrows andsequences of frames to help viewersunderstand the causal chains of motion that transform a driving forceinto mechanical movement. Afteridentifying the design principles, weimplemented them algorithmicallywithin interactive systems for generating cutaways, exploded views, andhow-things-work illustrations (seeFigures 2, 3, and 4).We applied a similar approach toidentify the design principles for depicting route maps that provide directions from one location to another1,3and destination maps that show multiple routes from all around a region toa single location (such as an airport ora popular restaurant).12 We analyzed avariety of such hand-drawn maps andfound they are often far more usefulthan computer-generated driving directions (available at sites like maps.bing.com and maps.google.com) because they emphasize roads, turningpoints, and local landmarks. Thesehand-designed maps significantly distort the distance, angle, and shape ofroads while eliminating many detailsthat would only serve to clutter theFigure 5. M apPoint screensh ot rep rinted wit h p ermissio n of Microsof t Corp.F igure 6. Google M apsFigure 5. A computer-generated route map rendered at a fixed scale does not depict (left) all the turns necessary for navigation. A handdesigned map (middle) emphasizes the turning points by exaggerating the lengths of short roads and simplifying the shape of roads. OurLineDrive system incorporates these design principles (right) into an automated map-design algorithm.

contributed articlesFigure 7. Unique M ediamap. Tufte23 pointed out that triptiksand subway maps similarly distortthe shape of routes and eliminateunnecessary detail. Hand-designeddestination maps include only themajor routes to a location rather thanall possible routes. These maps progressively increase the level of detail,showing only the highways far fromthe destination while including arterial roads and finally the residentialroads near the destination. Both routeand destination maps typically usemulti-perspective rendering in whichthe roads are drawn in top-down planview while important landmark buildings are drawn from a side view sotheir facades are visible.Although analyzing hand-designedvisualizations is often a good initialapproach for identifying design principles, this strategy also involves limitations. In some cases it may be tempting to form generative rules that aretoo specific and do not apply outsidethe range of analyzed examples. Inother cases the rules may be so generalit is unclear how to apply them to specific examples. Such difficulties oftenarise when the perceptual or cognitiverationale behind a particular visualtechnique is not clear. In the contextof route maps, for example, althoughour analysis revealed that mapmakersoften distort road length, angle, andshape, it was not immediately clearhow such distortions improved the perception and cognition of route maps.Similarly, we have found that oneof the challenges in analyzing handdesigned visualizations is to factorout differences due to artistic style.Designers may choose visual attributes (such as font type, color palette,and line width) for aesthetic reasonswhereby one font may simply looknicer than another to the designer. Although such aesthetic design choicesare important considerations, thegoal of our analysis is to determinehow the design choices improve theperception and cognition of the information, rather than how these choicesimprove aesthetics. The difficulty isthat these design choices often affectboth the aesthetics of the display andthe perception and cognition of theinformation; how to separate the twoeffects is not always clear.In light of these limitations andchallenges, we have found it is oftenuseful to connect our observationsand hypotheses from the analysis ofhand-designed examples with relevantwork from perception and cognitivepsychology. These connections serveto clarify the perceptual or cognitiverationale for the design principles.Prior work in perception and cognition. In some cases, prior researchin perception and cognition suggestsor formalizes the appropriate designprinciples; for example, cognitive psychologists have shown that peoplethink of routes as a sequence of turns25and that when following a route theexact length of a road is far less important than properly executing the turns.The topology of the route is more important than its absolute geometry.This insight helps explain why handdrawn maps often distort geometry—distance, angle, and shape of roads—to ensure that all roads and turningpoints are visible, but almost nevermodify the topology of the route.In this case, the prior research confirmed and formalized the perceptual/cognitive rationale for the visual techniques we first noticed when analyzing hand-drawn route maps. Based onthe resulting design principles, we developed LineDrive (http://vis.berkeley.edu/LineDrive), a fully automated system for rendering route maps in thestyle of hand-drawn maps.3 LineDrivehas been publicly accessible since October 2000, and surveys have shownthat for navigation tasks users strongly prefer LineDrive maps to computergenerated maps drawn at a fixed scale(see Figure 5).Researchers have also found thatnavigators familiar with a geographicarea (such as cab drivers) plan routeshierarchically.4 They first select thehighways necessary to get close to thedestination, then the arteries, and finally the residential streets. Such hierarchical planning corresponds tothe progressive increase in road detailwe first identified in hand-designeddestination maps. We recently applied this level-of-detail principle inconjunction with the distortion principles to build an automated systemfor generating destination maps.12 Asin LineDrive, we produced a map thatlooks hand-drawn but that eliminatesclutter while preserving the information necessary for anyone in thesurrounding region to reach the destination. Our destination maps areFigure 7. A hand-designed tourist map of San Francisco emphasizes semantically, visually, and structurally important landmarks, paths,districts, nodes, and edges, using multi-perspective rendering to ensure the facades of buildings are visible (left). Our tourist-map designsystem is based on these principles and similarly emphasizes the information most important for tourists in this map of San Francisco (right).april 2 0 1 1 vol. 5 4 n o. 4 c o m m u n ic at i o n s of t he acm65

contributed articlesavailable through the Bing Maps Website (http://vis.berkeley.edu/DestMap)(see Figure 6).We applied a similar approach toautomatically generating maps fortourists visiting a new city.8 Prior workon mental representations of cities16showed that people consider five mainelements: landmarks, paths, districts,nodes, and edges. However, a mapwith every instance of such elementswould be cluttered with excessive detail. The most-effective tourist mapsinclude only those elements that aresemantically meaningful (such as thehome of a well-known writer), visuallydistinctive (such as an oddly shapedor colored building), or placed in astructurally important location (suchas a building at a prominent intersection).22 After choosing the elements toinclude in the map, mapmakers usually apply a variety of cartographicgeneralization techniques, includingsimplification, displacement, deformation, and selection. Cognitive psychologists and cartographers studying the cognition of maps have shownsuch generalizations improve claritybecause they emphasize the most important map elements while preserving spatial relationships betweenthese elements.17Our tourist-map-design system isbased on these design principles. Input consists of a geometric model of acity, including streets, bodies of water,parks, and buildings (with textures).The system automatically determinesthe importance of map elements us-ing top-down Web-based informationextraction techniques to computesemantic importance and bottom-upvision-based image/geometry analysis to compute visual and structuralimportance. It then generates a mapthat emphasizes the most importantmap elements, using a combinationof multi-perspective rendering andcartographic generalization to highlight the important landmarks, paths,districts, nodes, and edges while deemphasizing less-important elements(see Figure 7).Experiments on perception andcognition. In some domains, new perception and cognition research is required to provide the rationale for thedesign principles. Working with cognitive psychologist Barbara Tversky,we developed a methodology for conducting human-subject experimentsto understand how people think aboutand communicate the informationwithin a domain. We first applied thismethodology to identify the designprinciples for creating assembly instructions for everyday objects (suchas furniture and toys).2,9 The experiments are conducted in three phases:Production. Participants create visualizations for a given domain. Inthe context of assembly instructions,they assembled a TV stand without instructions using only a photograph ofthe assembled stand as a guide. Theythen drew a set of instructions showing how to assemble it;Preference. Participants rate the effectiveness of the visualizations. Inthe assembly-instructions project, anew set of participants assembled theTV stand, without instructions. Theythen rated the quality of the instructions created by the first set of participants, redrawn to control for clarity,legibility, and aesthetics; andComprehension. Participants usethe ranked visualizations, and we testfor improvements in learning, comprehension, and decision making. Inthe assembly-instructions project, yetanother set of participants assembledthe TV stand, this time using the instructions rated in the preferencephase. Tests showed the highly ratedinstructions were easier to use andfollow; participants spent less timeassembling the TV stand and madefewer errors.Following these experiments, welook for commonalities in the highlyrated visualizations to identify thedesign principles. In the context ofassembly instructions, we identifiedthree main principles: (1) use a stepby-step sequence of diagrams showing one primary action in each diagram; (2) use guidelines and arrows todepict the actions required to fit partstogether; and (3) ensure that the partsadded in each step are visible. Our automated assembly-instruction-designsystem is based on these principles(see Figure 8). Tversky and Lee25 havestudied mental representations ofmaps using a similar methodology,where subjects first draw maps to familiar locations, then other subjectsrate the effectiveness of the maps.Figure 8. We asked subjects to assemble a TV stand and then create instructions for a novice explaining how to assemble it (left, middle).Analyzing hand-drawn instructions, we found that diagrammatic, step-by-step instructions using guidelines and arrows to indicate theactions required for assembly and providing good visibility for the attached parts are easiest to use and follow. Our system automaticallygenerates assembly instructions (right) based on these principles.66communi cations o f th e acm April 2 0 1 1 vol . 5 4 no. 4

contributed articlesStage 2. InstantiateDesign PrinciplesDesigning a visualization usually requires choosing visual properties orattributes for each element in thedisplay; for example, to create a routemap, the designer must choose attributes, including position, size, andorientation for each road, landmark,and label that appears in the map.Similarly, to create a cutaway illustration, the designer must choose howand where to cut each structure thatoccludes the target part. Because thereare many possible choices for each attribute, the design space of possiblevisualizations is usually quite large. Tobuild automated visualization designsystems, we treat the relevant designprinciples as guidelines for makingthese design decisions. The principleshelp us navigate through the designspace and obtain an effective design.Most design principles are statedas qualitative guidelines, rather thanas procedures we can directly instantiate in an automated design algorithm. The challenge is to transformsuch high-level principles into implementable algorithms.Design principles generally fallinto two categories: design rules andevaluation criteria. Design rules separate the design space into regions containing effective designs and thosecontaining inviable designs. They areessentially hard constraints in the design space. In creating route maps, forexample, designers commonly adjustthe turn angle to emphasize the orientation of the turn, to the left or tothe right. However, adjusting the turnangle so much that a left turn appearsto be a right turn or vice versa is unacceptable. This design rule puts a hardconstraint on how much designers areable to adjust the turn angle.Evaluation criteria quantify theeffectiveness of some aspect of thevisualization. We can assess the overall effectiveness of a visualization byconsidering a set of evaluation criteria covering all major aspects of thevisual design. In creating an explodedview, for instance, designers must balance two such criteria: part separationand compactness. A good explodedview separates the parts so all of themare visible, yet the visualization mustalso remain compact and maintain aThese principlesexplain howvisual techniquescan be used toeither emphasizeimportantinformation orde-emphasizeirrelevant detailsin the display.roughly square aspect ratio to makethe best use of available screen space.To quantify the overall effectivenessof an exploded view we measure thevisibility of each part, as well as thecompactness of the overall visualization. Similarly, in designing routemaps, designers must ensure that allroads are visible. To quantify this criterion, we compute the length of eachroad in the map and check the lengthis greater than some minimum visibility threshold. The number of roadslonger than the threshold length is aquantitative measure of the effectiveness of the map with respect to thiscriterion.Given a set of design rules andquantitative evaluation criteria, wecan use procedural techniques tobuild an automated visualization design system; for example, our systemfor designing cutaways and explodedviews is driven exclusively by procedural techniques. In this case, we encode the design rules as a decisiontree describing how to cut or explodeaway occluding parts based on theirgeometry. Another approach is to consider visualization design as an energy-minimizing optimization problem.In this case, we treat the design rulesas hard constraints that define theboundaries of the design space andthe evaluation criteria as soft constraints that guide the system to theoptimal visualization. While this optimization-based approach is general,we have found it essential to develop aset of design rules and evaluation criteria that sufficiently limit the designspace so it is feasible to complete theoptimization. Both LineDrive and ourassembly-instruction design systemuse such an energy-minimizing optimization.Stage 3. Evaluate Design PrinciplesThe final stage of our approach is tomeasure the usefulness of the visualizations produced by our automateddesign systems. We consider severalsuch measures, including feedbackfrom users in the form of qualitativeinterviews and quantitative usage statistics. In some cases, we have alsoconducted more-formal user studiesto check how well the visualizationsimprove information processing, communication, and decision making.april 2 0 1 1 vol. 5 4 n o. 4 c o m m u n ic at i o n s of t he acm67

contributed articlesUser feedback. We find it is criticalto involve users early on and conductqualitative interviews and surveys tocheck their overall impressions of thevisualizations produced by our systems. Such feedback is essential foridentifying problems and ensuringour design principles and the visualizations converge on effective designs.The interviews and surveys providehigh-level checks of the effectivenessof our design principles and allow usto tweak the principles when not quiteright; for example, early on buildingLineDrive, we asked users to rate handcrafted prototype route-map designs,finding that 79 out of 90 respondentspreferred the distorted LineDrive prototypes to maps drawn to scale1 andconfirming that users thought the distorted maps were useful. Continualfeedback and evaluation yields moreeffective algorithms and tools.Another approach is to release thevisualization on the Web, then checkusage statistics; for example, at itspeak, LineDrive was serving more than750,000 maps per day and became thedefault route-mapping style for MapBlast, an early Web-based provider ofroute maps. Such public feedback isa strong test of effectiveness, as ineffective solutions are quickly rejected.We also recognize that u

Design Principles for Visual Communication key insights Design principles connect the visual design of a visualization with the viewer’s perception and cognition of the underlying information the visualization is meant to convey. i dentifying and formulating good design p