<-- Advertise here.

a visualization that allows users to quickly identify the most significant differences in content between 2 lists of Web pages, meant to help comparing the effects of censorship policies on search engines. one interface simply compares the image thumbnails of 2 identical search queries by putting them next to each other. the textual visualization shows the individual tag clouds of the 50 terms that are most particular to each query's results.

while the goal seems quite intriguing, the visualizations seem to require extensive visual comparisons. any ideas how the authors could make the interface more efficient?

[link: uic.edu|thnkx Su]





How about Many Eyes' Comparable Tag Cloud? For the text part it would be very helpful if the compared words were next to each other...

Thu 24 Jul 2008 at 12:31 AM

It's interesting enough that a more comprehensive version would be welcome. As is, the fact that it returns differing results for a search on one country (USA, USA, cf. USA, China) makes it susspect.

I did a search on 'hateful' with both countries set to USA. The left side returned a very different set than the right.

Thu 24 Jul 2008 at 10:55 AM

Hmm, dp, I think you may be wrong. I tried the same thing -- searched "hateful" with the US selected for each -- and I got zero results. I think you need to refresh your browser. Before I refreshed, I found that old results persisted -- that is, I had results on one side from a previous search for Germany, and new results for the new searhc for the US on the other side. Try refreshing your browser and running it again. I think you'll find it less suspect.

Fri 25 Jul 2008 at 6:20 AM

the interface should at least let the viewer go "n^2" on the results by letting them select matches and remove them- ala scraping off the crust. i think there is some cog results that show people can hold up to a 9-node fully connected graph in memory. beyond that is impossible without external symbols. (number may be off)

Fri 25 Jul 2008 at 3:07 PM

I don't believe it's anything revolutionary, but arranging the words relative to an axis shared by both might aid in the identification of common words. Such an axis might run from top to bottom in the center of the image, and represent continuum of words ordered alphabetically. The keywords collected would be distributed across this axis, the left side reserved for the first search engine, and the right for the other. To further associate the two sides of the visualization, the left side might be right-aligned against the center, and the right side left-aligned similarly.

Sun 10 Aug 2008 at 9:15 PM
Commenting has been temporarily disabled.