I’ve been analyzing the content of blogs lately, looking for patterns. It’s a huge amount of data, which makes for some tricky technical problems. Finally, tonight, thanks to some help from friends and the Large Graph Layout package, I’ve finally got some results. And they’re stunning. Ladies and gentlemen, the blogosphere:

And, for fun, let’s zoom in one one of those small splotches:

posted 2006-07-26T02:53:22 #
Interesting. So when do we get more details? How are you determining the proximity of concepts? And what are those three giant splotches in the middle?
posted by Scott Reynen
on 2006-07-26T03:25:52 #
Bush, academia, and law, I think. As I said, this is just the very first result; hopefully I can share more interesting stuff later.
posted by Aaron Swartz
on 2006-07-26T03:27:29 #
Aaron - where did you get your dataset from?
posted by Kunal
on 2006-07-26T04:18:19 #
Kunal, you could work based off of Alexa’s platform and just fiddle with linking.
posted by Jeremy Dunck
on 2006-07-26T04:36:22 #
The above graph appears to be a scale-free network (not surprising, really). Aaron, have you analyzed it with regard to its scale-free properties?
posted by Toby
on 2006-07-26T14:14:57 #
Not surprisingly, it looks an awful lot like the most other small networks. Interesting to note are the patterns supported by the Power Law.
posted by greg
on 2006-07-26T16:21:09 #
No chance to get an anti-aliased version of that graph to see a bit more of the structure? This is totally wicked.
posted by joe
on 2006-07-26T18:06:12 #
…it would be cool to lay it out inside a 3D sphere… then it would really be the blogosphere.
posted by joe
on 2006-07-26T18:07:32 #
That dataset of yours probably doesn’t have an open license? ;)
posted by Chris Laux
on 2006-07-26T18:36:43 #
I don’t believe Reddit has an API yet.
posted by Kunal Anand
on 2006-07-28T02:28:28 #
Very appealing; I would like to get more information on how you decomposed the data, and how you produced the layout.
posted by Matthieu Latapy
on 2007-07-25T22:25:46 #
Very appealing; I would like to get more information on how you decomposed the data, and how you produced the layout.
posted by Matthieu Latapy
on 2007-07-25T22:26:22 #
Letters to the editor are printed at the discretion of the proprietor. They may be edited for length and content.
(You can also send your letters by email. If you choose to do so, please note if you're willing to make your letter available for publication.)
Interesting. So when do we get more details? How are you determining the proximity of concepts? And what are those three giant splotches in the middle?
posted by Scott Reynen on 2006-07-26T03:25:52 #
Bush, academia, and law, I think. As I said, this is just the very first result; hopefully I can share more interesting stuff later.
posted by Aaron Swartz on 2006-07-26T03:27:29 #
Aaron - where did you get your dataset from?
posted by Kunal on 2006-07-26T04:18:19 #
Kunal, you could work based off of Alexa’s platform and just fiddle with linking.
posted by Jeremy Dunck on 2006-07-26T04:36:22 #
The data is from http://feeds.reddit.com/.
posted by Aaron Swartz on 2006-07-26T04:38:35 #
The above graph appears to be a scale-free network (not surprising, really). Aaron, have you analyzed it with regard to its scale-free properties?
posted by Toby on 2006-07-26T14:14:57 #
Not surprisingly, it looks an awful lot like the most other small networks. Interesting to note are the patterns supported by the Power Law.
posted by greg on 2006-07-26T16:21:09 #
No chance to get an anti-aliased version of that graph to see a bit more of the structure? This is totally wicked.
posted by joe on 2006-07-26T18:06:12 #
…it would be cool to lay it out inside a 3D sphere… then it would really be the blogosphere.
posted by joe on 2006-07-26T18:07:32 #
That dataset of yours probably doesn’t have an open license? ;)
posted by Chris Laux on 2006-07-26T18:36:43 #
I don’t believe Reddit has an API yet.
posted by Kunal Anand on 2006-07-28T02:28:28 #
Very appealing; I would like to get more information on how you decomposed the data, and how you produced the layout.
posted by Matthieu Latapy on 2007-07-25T22:25:46 #
Very appealing; I would like to get more information on how you decomposed the data, and how you produced the layout.
posted by Matthieu Latapy on 2007-07-25T22:26:22 #
Letters to the editor are printed at the discretion of the proprietor. They may be edited for length and content.
Add yours
(You can also send your letters by email. If you choose to do so, please note if you're willing to make your letter available for publication.)