Over at DataVisualization.CH, Interactive Comments From Digg.com was posted the other day. [Note: The site loads a bit slowly sometimes. Be patient, it is worth the wait.]

An Overhead View

Alternate Views

Commentry is an interactive 3d visualization of comments on a popular digg story. Digg’s rating system combined with word counts allows us to see the status as well as health of a story. Heated debates can easily be found as well as popular reactions(good or bad) made by a given user.

I was intrigued by the visualization and made the following comment over at the site:

Regarding the usage of a three dimensional Pie Chart, it is an interesting question. I have trouble understanding what I am looking at in the visualization. I’m assuming blue is positive rating and red is negative rating; however, blue has negative numbers and red positive numbers. It must be the other way around.

I’m very interested in techniques that try to visualize unstructured text. I hope I’m understanding what is represented. Each wedge represents the % of words in a particular comment (or a comment and the replies to the comment) as part of a whole discussion. The length of the wedge gives an indication of the volume of replies. Lastly, there is a vertical component that shows rating.

If that is true, than the widest, longest, highest wedge would seemingly be where most of the highly-rated, active discussion is occurring.

The 3-d representation always will have a scaling problem when viewed in 2-d. Additionally, there are 3 dimensions, but just one text label. It is not totally clear what that number represents. I think it is the positive/negative quality. As I looked at the visualization though, I kept thinking it should represent the length of the wedges.

For me, I’m not positive my understanding of the discussion is enhanced by the visual. It would be interesting to know what question the visualization is addressing. What would someone do with this information?

It is an interesting project. Visualizing text is still in its infancy. All experimentation will advance best practices.

Thanks for bringing this to our attention. Sorry for my rambling comment.

Both the post author and the project developer have responded. Go over, take a look, and join the discussion. There is some really nice material over at DataVisualizations.CH.

Comments