Collaborative Visual Analysis

From CS 294-10 Visualization Sp10

Jump to: navigation, search

Lecture on April 21, 2010




  • Voyagers and Voyeurs: Supporting Asynchronous Collaborative Information Visualization. Heer et al.(html)
  • Many Eyes: A Site for Visualization at Internet Scale.Viégas et al. (pdf)

Optional Readings

  • Design Considerations for Collaborative Visual Analytics. Heer and Agrawala. (html)

Stephen Chu - Apr 26, 2010 01:24:05 pm

One difficulty in creating visualizations is figuring out how the visualization will be used/perceived by others. I like the idea of using ManyEyes/CommentSpace as a platform for user testing visualization techniques. Sites like ManyEyes make it easier and quicker to expose visualizations to a large audience, and tracking social collaboration will be very helpful in gauging a visualization's performance. These platforms give us a new way of studying how people use visualizations, what they find important, how they navigate to find answers to their questions, etc. An example of where social data collaboration would have been interesting/helpful is Enabling even simple text comments on the stack page would have cleared some confusion (or perhaps created more) as well as add an avenue for direct feedback to the creators.

Jon Barron - Apr 26, 2010 07:21:20 pm


It seems that the most important issue that this paper does not provide a satisfactory, systematic solution for is how an annotation in one view of the data is visible from other views of the data. The idea of "doubly-linked" discussion kind of addresses this issue, but it seems to go much deeper. Given, for example, the annotation in Figure 2a where Nevada is highlighted, on the plot showing gender distribution, how should that annotation appear when visualizing, for example, a 1D histogram of the gender distributions of states? The data on which they demonstrate collaboration seems to have the property that there really is only one view (the "voyager" view, typically) through which things can be seen. How can annotations generalize to other views?

Connected with this concern is the trend they show, where the vast majority of users prefer annotating with non-machine-readible annotations (pencil, arrows, lines) as opposed to annotations that have easily defineable meaning (rectangles, ovals, text --- somewhat). One could imagine transforming a rectangle on a map into some other symbol on a histogram, but how do text and arrows get transformed? Or cartoons of stock brokers?

Jaeyoung Choi - Apr 28, 2010 01:29:45 am

Many Eyes: To embrace non-programmers to a new service which requires social collaboration, instant usability, and immediate accessibility should be seriously considered. Many design choices to fulfill these were mentioned in the paper, but I wonder why they gave up to add some more features (such as allowing to import XML-formatted data) and let the end users to choose.

Soundcloud allows artists and music professionals to upload/receive/distribute their music and listeners to put their comment on the waveform representation of the music at any point wherever they feel like. This is very limited when compared to Many Eyes, but I think this is still very cool as you can easily catch how people's impression on each part of the music changes as the music play.

Arpad Kovacs - Apr 28, 2010 03:00:12 pm

The articles have two authors in common, and it seems that both the and the Many Eyes web sites are attempts by IBM to develop ways for asynchronous collaborative visualization with an emphasis on social collaboration. While the first article is more concerned about visualization and annotations, the second article's focus is on easy usability and serving a large audience provided by the Internet.

Heer: Supporting Asynchronous Collaborative Information Visualization I was surprised by how interactive visualization and visualization-related comments could foster social interaction in the way described in this paper. The prototype system truly facilitated this interaction. I especially liked the bookmark trail that shows multiple views (storytelling) within a single comment. This system could be used not only to annotate and comment on all kinds of static graphics, but also to display progressions, changes and trends in a flipbook animation-like fashion, and complement and enhance static image based visualization techniques.

Viegas: A Site for Visualization at Internet Scale A key concern of this experiment was the large size of the Internet audience and the immediate usability of the tools. Another challenge was simplifying data uploading for nontechnical users. I thought that automatically generating areas on the bottom of the charts, which allow the user to select different dimensions and data sets, was an innovative idea. My feeling is that the current site is more like a sandbox to test different visualization techniques and gain an understanding about how average (not computer or data savvy) users utilize visualization. Right now there are very few channels for such user feedback, because visualization tools usually do not provide metrics about the usage of their tools/features or allow users to comment on them. Therefore this experiment seems to be a worthwhile attempt to widen the knowledge about user expectations and actual usage of visualization tools and techniques.

Zev Winkelman - May 02, 2010 11:32:48 pm

Heer et al. - I thought the most interesting part of this work was the doubly linked discussions. The technique presented by the authors, of treating filter parameters and view parameters differently, was very interesting to me, but I wonder how the indexing will scale when the discussion become less sparse and more granular. I can imagine some extreme cases where the size of the index data could grow quite large.

Viégas et al. - I would like the Many Eyes project ( and others like it ) much better if they offered a non web based version and/or open source like some of the other software we've seen. I wonder whether the sharing of data, especially with a login, is a significant barrier to greater use.

Heer and Agrawala - I like the breakdown of collaboration as collocated, synchronous and asynchronous, and the references to Benkler's work on modularity, granularity and the cost of integration. I was also particularly interested in the highlighting of the difficulty under conditions of a competitive model of sensemaking to integrate the end results and present them.

Prahalika Reddy - May 12, 2010 08:31:19 am

The concept of collaborative visual analysis is an interesting one. More often that not, I've thought of visualizations as individual works, instead of collaborative ones. However, adding the power of collaboration to the power of visualizing something only makes it better. Collaborative visualizations are oftentimes very effective, just as many collaborative efforts are.

I really like the tool because of the amount of information it provides to users. Incorporating the Job Voyager visualization only makes it more effective and allowing users to add in their thoughts helps everyone get more out of the data.

Shimul Sachdeva - May 13, 2010 05:52:10 am

The demo showed in class was interesting. It is a great tool to get good feedback on visualizations, although I wonder if there is a way to filter out inappropriate/irrelevant comments that tend to go off-tangent. I especially like the ability to highlight a certain section and comment on it - highly interactive!

The concept of collaborating visualizations is an efficient way to bring information together and thus helps identify important details that may go unnoticed otherwise. ManyEyes seems a little loosely defined in terms of its collections, but if its database is filtered correctly, it can potentially be a powerful tool. In addition, if there are options of obtaining data sets in desired formats and a way to certify whether the data is legitimate, that would be helpful.

DEVise lacks visual appeal and given it is a collaborating tool for visualizations, it is unimpressive.

[add comment]
Personal tools