Visualization Mini-Workshop

Please join us on Friday May 12 from 1:00-4:30pm in Soda 306 for a mini-workshop on recent research in Visualization techniques.

Visualizations are displays that are designed to convey information. They capitalize on human facilities for processing visual information and when well designed they can improve comprehension, memory, inference, and decision making. Yet, designing visualizations remains a difficult and time-consuming task. Research on visualization techniques is aimed at developing principles, algorithms and methodologies that will make it easier to produce effective visual displays.

On May 12 we will be hosting talks from a number of visualization researchers.

Schedule

1:00-1:05: Maneesh Agrawala - Welcome and Introductions

1:05-1:25: Jeff Heer - Multi-Scale Banking to 45 Degrees

1:30-2:20: Jock Mackinlay & Chris Stolte - Visual Interfaces for Databases

2:30-2:55: Steve Drucker - Novel Interfaces for Working with Digital Photo Collections

3:00-3:25: Tamara Munzner - An Evaluation of Pan&Zoom and Rubber Sheet Navigation with and without an Overview

3:30-3:55: Stephen Few - Bridging the InfoVis Chasm

4:00-4:25: Marti Hearst - Tailored vs. Off-the-Shelf Infoviz Solutions

Talk Abstracts

Multi-Scale Banking to 45 Degrees

Jeffrey Heer

In his text Visualizing Data, William Cleveland demonstrates how the aspect ratio of a line chart can affect an analyst’s perception of trends in the data. Cleveland proposes an optimization technique for computing the aspect ratio such that the average absolute orientation of line segments in the chart is equal to 45 degrees. This technique, called banking to 45°, is designed to maximize the discriminability of the orientations of the line segments in the chart. In this talk, we revisit this classic result and describe two new extensions. First, we propose alternate optimization criteria designed to further improve the visual perception of line segment orientations. Second, we develop multi-scale banking, a technique that combines spectral analysis with banking to 45°. Our technique automatically identifies trends at various frequency scales and then generates a banked chart for each of these scales. We demonstrate the utility of our techniques in a range of visualization tools and analysis examples.

Novel interfaces for working with digital photo collections

Steven Drucker

Managing the ever growing personal collections of digital photographs is becoming an increasing challenge. This talk will highlight three novel interfaces that explore different aspects of photo collection management on rich clients. All the interfaces use animation to help the user maintain context as they explore the collection. The first interface – MediaFrame – is a system designed to help users create annotations and add metadata to photos by combining automatic assistance, smooth motion, and simple, consistent tools. The second – TimeQuilt – explores how to scale and browse larger collections of photos without extensive annotation. Finally, PhotoTriage, is aimed specifically at the task of quickly sorting recent photos into different categories.

Visual Interfaces for Databases

Jock Mackinlay & Chris Stolte

Although visual interfaces and databases are two of the success stories of the computer revolution, their synergy to date has been modest, probably because visual interfaces have focused on human capabilities while databases have focused on efficient query processing. The success of visual interfaces started with the GUI (Graphical User Interface), which supplanted the command line interface by exploiting the power of the human visual motor system. Given advances in graphics hardware in the mid 1980s, research started on Visualization, the use of interactive, visual representations of data to amplify cognition. We will briefly describe Mackinlay’s dissertation, which formalized Jacques Bertin’s design theory, adding psychophysical data, resulting in a system that could automatically design graphical presentations. The 1990s were a fertile time for Visualization research, culminating in a book co-authored by Mackinlay that included a formal reference model for describing visualization systems. However, this research had little impact on databases even though queries are essentially a command line interaction. Rather, the success story of databases started with the invention of relational databases that supported efficient transactions. The 1980s saw a huge effort to rework our institutions to use computers to manage our vital data ranging from our birth statistics to the legacy that we leave our children. In the 1990s we developed multi-dimensional databases to create data warehouses for the efficient analysis of this data. However, analysis and exploration place significant demands on the interfaces to these databases, which might be addressed with visual interfaces for databases. Because of the size of the data sets, dense graphical representations are more effective for exploration than spreadsheets and charts. Furthermore, because of the exploratory nature of the analysis, it must be possible for the analysts to change visualizations rapidly as they pursue a cycle involving first hypothesis and then experimentation. In this talk we describe an interface for exploring large multi-dimensional databases that extends the well-known Pivot Table interface. The novel features include an interface for constructing visual specifications of table-based graphical displays and the ability to generate a precise set of relational queries from the visual specifications. The visual specifications can be rapidly and incrementally developed, giving the analyst visual feedback as they construct complex queries and visualizations.

An Evaluation of Pan&Zoom and Rubber Sheet Navigation with and without an Overview

Tamara Munzner

We present a study that evaluates conventional Pan and Zoom Navigation and Rubber Sheet Navigation, a rectilinear Focus+Context technique. Each of the two navigation techniques was evaluated both with and without an overview. All interfaces guaranteed that regions of interest would remain visible, at least as a compressed landmark, independent of navigation actions. Interfaces implementing these techniques were used by 40 subjects to perform a task that involved navigating a large hierarchical tree dataset and making topological comparisons between nodes in the tree. Our results show that Pan and Zoom Navigation was significantly faster and required less mental effort than Rubber Sheet Navigation, independent of the presence or absence of an overview. Also, overviews did not appear to improve performance, but were still perceived as beneficial by users. We discuss the implications of our task and guaranteed visibility on the results and the limitations of our study, and we propose preliminary design guidelines and recommendations for future work.

Bridging the InfoVis Chasm: From the Theoretical to the Practical, the Innovative to What Really Works, and the Cool to What’s Really Needed

Stephen Few

We need to build a better bridge between the finest work that is being done by the information visualization research and development community and the commercial software that is desperately needed by the business community. To span his chasm, the infovis community must learn to distinguish flights of fancy from practical solutions that work in the real world. What must researchers and developers know and do to produce commercial software that will excite business users, meet their most critical needs, and fully engage their brains in the sense-making process? In this presentation, Stephen Few, a leader in the integration of information visualization and business intelligence, will describe what’s needed and what must be done to produce it.

Bios

Jeffrey Heer

Jeffrey Heer is a 4th year Ph.D. student in the Computer Science Division at U.C. Berkeley, working with the Group for User Interface Research and Berkeley Institute of Design. Jeff is also an alumnus of the Xerox Palo Alto Research Center and more recently of Tableau Software, an information visualization company spun out of Stanford. Jeff's research interests lie in Human-Computer Interaction, where he has worked on theory, tools, and techniques for Information Visualization and on the design and implementation of various Ubiquitous Computing systems.

Jock Mackinlay

Jock Mackinlay received his PhD in computer science from Stanford University, where he pioneered the automatic design of graphical presentations of relational information. He joined Xerox PARC in 1986, where he collaborated with the User Interface Research Group to develop many novel applications of computer graphics for information access and to coin the term "Information Visualization". Much of the fruits of this research can be found in his book, Readings in Information Visualization: Using Vision to Think (Morgan Kauffman, written and edited with Stuart K. Card and Ben Shneiderman). He holds numerous patents in user interfaces and is a member of the editorial board of IEEE Computer Graphics & Applications. In 2004, he joined Tableau Software as Director of Visual Analysis.

Chris Stolte

Chris Stolte is a co-founder of Tableau Software and Vice President of Engineering. Chris has been researching the visual analysis and exploration of databases for the last eight years and his research has resulted in ten research publications and two large-scale visualization systems. Chris was also CTO and co-founder of BeeLine Systems, a visualization software company that developed a revolutionary map rendering system. BeeLine was purchased by Vicinity Corporation (NASDAQ: VCNT), and its products are currently used to generate over a million maps a day. Chris is a co-inventor on five software patents related to information visualization. He holds a Ph.D. in Computer Science from Stanford University, and a B.S. in Computer Science from Simon Fraser University.

Steven Drucker

Dr. Steven M. Drucker recently joined the LiveLabs Group at Microsoft as a Senior Scientist. He is working on user interaction and information visualization for web based projects. Previously, he was the lead researcher for the Next Media Research Group in Microsoft Research for 6 years where he has looked at how the addition of user interaction transforms conventional media. He is particularly interested in database visualization for consumers or where art meets technology for user interfaces. While in the group, he has filed 35 patents on technologies as diverse as remotely operated personal video recorders, spectator oriented gaming, and new visualization techniques for media databases as well as published papers on information visualization and management.

Tamara Munzner

Tamara Munzner received a BS in 1991 and a PhD in 2000 from Stanford. She was on the technical staff of The Geometry Center, a mathematical visualization research group at the University of Minnesota, from 1991 to 1995. From 2000 to 2002 she was a research scientist at the Compaq Systems Research Center in California. She has been an assistant professor in the University of British Columbia Department of Computer Science since 2002. Her current research interests are information visualization, graph drawing, and dimensionality reduction. She was the IEEE Symposium on Information Visualization (InfoVis) Program/Papers Co-Chair in 2003 and 2004.

Stephen Few

Stephen Few has worked for 24 years as an IT innovator, consultant, and educator. Today, as Principal of the consultancy Perceptual Edge, Stephen focuses on data visualization for analyzing and communicating quantitative business information. He provides consulting and training services, writes the monthly data visualization column for the Business Intelligence Network (www.B-EYE-NETWORK.com), speaks frequently at conferences like TDWI and DAMA, and teaches in the MBA program at the University of California, Berkeley. He is also the author of two books: Show Me the Numbers: Designing Tables and Graphs to Enlighten and a new book entitled Information Dashboard Design: The Effective Visual Communication of Data.