News Papers Covered Design Study Ware, Chapter 1: Foundation for a - PowerPoint PPT Presentation

News Papers Covered Design Study Ware, Chapter 1: Foundation for a Science of Data Visualization Lecture 2: Design Studies ◮ questions Cluster and Calendar based Visualization of Time Series Data. Information Visualization ◮ one question per reading, up to max of 5 Jarke J. van Wijk and Edward R. van Selow ◮ plain (ASCII) text not Word/PDF/etc Proc. InfoVis 99, pp 4-9 CPSC 533C, Fall 2007 ◮ describe task ◮ reserve books http://www.win.tue.nl/˜vanwijk/clv.pdf ◮ justify solution ◮ in CS reading room (ICICS/CS 2nd floor) Using Multilevel Call Matrices in Large Software Projects. Tamara Munzner ◮ www.cs.ubc.ca/local/reading/about Frank van Ham ◮ refine until satisfied ◮ 24-hr fob access possible Proc. InfoVis 2003, pp 227-232 UBC Computer Science http://www.win.tue.nl/˜fvham/DL/callmatrix.pdf ◮ proxy server ◮ note instructions on web page for accessing IEEE DL Constellation: Linguistic Semantic Networks 12 September 2006 Tamara Munzner material Interactive Visualization of Large Graphs and Networks (PhD thesis) Chapter 5, Stanford University, 2000, pp 87-122 http://graphics.stanford.edu/papers/munzner thesis Design Study Definition Cluster-Calendar, van Wijk 3D Time-series Data Hierarchical Clustering ◮ 3D extrusion pretty but not useful ◮ start with all M day patterns ◮ data: N pairs of (value, time) ◮ daily, weekly patterns hard to see ◮ compute mutual differences, merge most similar: M-1 ◮ N large: 50K Design study papers explore the choices made when applying ◮ continue up to 1 root cluster ◮ tasks ◮ result: binary hierarchy of clusters infovis techniques in an application area, for example relating the visual encodings and interaction techniques to the ◮ find standard day patterns ◮ choice of distance metrics requirements of the target task. Although a limited amount of ◮ find how patterns distributed over year, week, season ◮ dendrogram display common application domain background information can be useful to ◮ find outliers from standard daily patterns ◮ but shows structure of hierarchy, not time distribution provide a framing context in which to discuss the specifics of ◮ want overview first, then detail on demand the target task, the primary focus of the case study must be the ◮ possibilities infovis content. Describing new techniques and algorithms ◮ predictive mathematical models developed to solve the target problem will strengthen a design ◮ details lost, multiscale not addressed study paper, but the requirements for novelty are less stringent ◮ scale-space approaches (wavelet, fourier, fractal) than in a Technique paper. ◮ hard to interpret, known scales lost [InfoVis03 CFP , infovis.org/infovis2003/CFP] ◮ 3D mountain: x hours, y value, z days ◮ excellent example, emulate for project writeups! [van Wijk and van Selow, Cluster and Calender based Visualization of Time Series [van Wijk and van Selow, Cluster and Calender based Visualization of Time Series Data, InfoVis99, http://www.win.tue.nl/˜vanwijk/clv.pdf] Data, InfoVis99, http://www.win.tue.nl/˜vanwijk/clv.pdf] Link Clusters and Calendar Power Consumption Lessons Lessons ◮ 2D linked clusters-calendars shows patterns ◮ number of employees: ◮ office hours, fridays in/and summer, school break ◮ derived space: clusters ◮ derived space: clusters ◮ weekend/holidays, post-holiday, santa claus ◮ visual representation of time: calendar ◮ visual representation of time: calendar ◮ linked display ◮ linked display ◮ interactive exploration ◮ interactive exploration ◮ clear task analysis guided choices ◮ clear task analysis guided choices ◮ reject standard 3D extrusion ◮ reject standard 3D extrusion ◮ reject standard dendrogram ◮ reject standard dendrogram ◮ critique ◮ critique ◮ color choice not so discriminable ◮ especially legend [van Wijk and van Selow, Cluster and Calender based Visualization of Time Series Data, InfoVis99, http://www.win.tue.nl/˜vanwijk/clv.pdf] [van Wijk and van Selow, Cluster and Calender based Visualization of Time Series Data, InfoVis99, http://www.win.tue.nl/˜vanwijk/clv.pdf] Multilevel Call Matrices, van Ham Matrices Zooming Additional Encoding ◮ uniform, recursive, stable ◮ abstraction levels color: color: transparency: ◮ subdivide by ◮ large software project, implementation vs. spec call allowed local region call density ◮ link matrix vs. node network total component count visible subcomponent count by spec closest red matrix force-directed layered subset ◮ linear interpolation plus crossfade ◮ trajectories: will read van Wijk 03 in nav lecture ◮ histograms: size distribution [van Ham, Using Multilevel Call Matrices in Large Software Projects. InfoVis03 [van Ham, Using Multilevel Call Matrices in Large Software Projects. InfoVis03 http://www.win.tue.nl/˜fvham/DL/callmatrix.pdf] http://www.win.tue.nl/˜fvham/DL/callmatrix.pdf] [van Ham, Using Multilevel Call Matrices in Large Software Projects. InfoVis03 [van Ham, Using Multilevel Call Matrices in Large Software Projects. InfoVis03 http://www.win.tue.nl/˜fvham/DL/callmatrix.pdf] http://www.win.tue.nl/˜fvham/DL/callmatrix.pdf]

Critique Linguistic Networks, Munzner Semantic Network Path Query ◮ data: MindNet query results ◮ definition graph ◮ dictionary entry sentence ◮ tasks succesfully supported ◮ best N paths between two words ◮ definition graphs used as building blocks ◮ visual categorization ◮ nodes: word senses ◮ words on path itself ◮ unify shared words ◮ links: relation types ◮ i.e. libraries with mostly incoming calls ◮ large network ◮ previous summary shown to be incomplete ◮ spotting unwanted calls ◮ millions of nodes ◮ definition graphs used in computation ◮ determining component dependencies ◮ grammar checking now, translation future ◮ limitations ◮ global structure known: dense ◮ ordering unsolved ◮ probes return local info ◮ strict layering, not true in many apps [Munzner, Interactive Visualization of Large Graphs and Networks (PhD thesis), Stanford University, 2000, http://graphics.stanford.edu/papers/munzner thesis] Task: Plausibility Checking Top 10 Paths Kangaroo → Tail Goal Constellation Video ◮ create a unified view of relationships between paths ◮ paths ordered by computed plausibility and definition graphs ◮ researcher hand-checks results ◮ shared words are key ◮ high-ranking paths believable? ◮ thousands of words (not millions) ◮ believable paths high-ranked? ◮ special purpose algorithm debugging tools ◮ are stop words all filtered out? ◮ not understand structure of English Traditional Layout Information Visualization Approach Constellation Semantic Layout Selective Emphasis ◮ avoid crossings ◮ spatial position is strongest perceptual cue ◮ highlight sets of boxes and edges ◮ encode domain specific attribute ◮ interaction ◮ reason: avoid false attachments ◮ additional perceptual channels ◮ plausibility gradient ◮ novel layout algorithm ◮ avoid perception of false attachments ◮ paths as backbone, definition graphs attached ◮ curvilinear grid ◮ iterative design for maximum semantics with reasonable information density ◮ allow crossings for long-distance proxy links ambiguity artifact salience Hidden State Single vs. Multiple Word Instances Information Density Information Density ◮ early prototype: poor ◮ design tradeoff with visual salience ◮ avoid hidden state ◮ change salience instead of toggle drawing ◮ why? closed world assumption ◮ implicit assumption: if not visible, doesn’t exist ◮ easy to forget previous actions ◮ draw false negative conclusions

Information Density Task-Oriented Design Readings For Next Time ◮ grid adjustment A Review and Taxonomy of Distortion-Oriented Presentation Techniques. Y.K. Leung and M.D. Apperley, ACM Transactions on Computer-Human Interaction, Vol. 1, No. 2, June 1994, pp. 126-160. [http://www.ai.mit.edu/people/jimmylin/papers/Leung94.pdf] A Fisheye Follow-up: Further Reflection on Focus + Context. George W. Furnas. SIGCHI 2006. The Hyperbolic Browser: A Focus + Context Technique for Visualizing Large Hierarchies. John Lamping and Ramana Rao, Proc SIGCHI ’95. [http://citeseer.nj.nec.com/lamping95focuscontext.html] TreeJuxtaposer: Scalable Tree Comparison using Focus+Context with Guaranteed ◮ task-specific methods Visibility. Munzner, Guimbretiere, Tasiran, Zhang, and Zhou. SIGGRAPH 2003. [http://www.cs.ubc.ca/˜tmm/papers/tj/] SpaceTree: Supporting Exploration in Large Node Link Tree, Design Evolution and Empirical Evaluation. Catherine Plaisant, Jesse Grosjean, and Ben B. Bederson. Proc. InfoVis 2002. ftp://ftp.cs.umd.edu/pub/hcil/Reports-Abstracts-Bibliography/2002-05html/2002-05.pdf