One Big Library.

At work I've been working with the amazing Starlight information visualization environment (to which we have brief trial access as part of potential grant work). I'm just beginning to understand how to get the most out of the system, so I'm only scratching the surface so far. Still, I'm already amazed by its capabilities.

To learn it I've been feeding it data from unalog using unalog's built-in rss and mods feeds. Starting with a set of 3000 bookmark entries, I've been able to produce some really unintelligible imagery (c.f. line about how "easy tools for making [music, maps, software] makes it easy for more people to make bad [music, maps, software]"). But, after getting more of a feel for it, a few interesting images have turned out.

This is the 3D topic view of a selected subset of records with the tag "python". The sphere is like a bar chart in three dimensions, but with links between tags used together on unalog bookmarks. The links are also color-coded by date using the rainbow spectrum.

This shows (where you can read it :) relationships between tags used at unalog, their relative frequencies, and to some extent, how all of that has evolved over a brief time interval. Within the Starlight console app, you can navigate this 3D space as you would expect to be able to.

The next image is the "link array" view, which relates multiple dimensions to each other. At top is a linear timeline, again color-coded using the rainbow. The plane below that is tags; below that are entry titles; at bottom are all entries. This view can also be navigated in three dimensions in the console app.

The unalog data subset here are from my (dchud's) unalog entries over the past seven months, taken straight from unalog's mods feed. All of the entries have thin/dimmed links between dates, tags, and their titles. Four are highlighted that have the tag "iceland"... one is from before our trip to Iceland last summer, three are from after. You can see this from the timeline at the top, which shows a clear two-week gap while we were away touring around the ringroad.

If you can make out the text, you'll see the names of some of our favorite stops, including a great record store and our favorite vegetarian restaurant. :)

There is a lot to critique here, particularly how the two-dimensional planar representation of titles and entries adds information that isn't useful. Where the planar representation is extended into a third dimension with bars of varying height (like in the tag plane, second from top), or color-coded (like in the bookmark entry plane, at bottom), useful context is added back in, even if it causes a cognitive processing delay.

There is *much* more to this visualization application than just what I can show here... I'm looking forward to doing a lot more with it, and hope to show more soon.

In the meantime if any of the unalog regulars wants to see their stuff in one of these images, let me know and I'll work something up. I need the practice!