Wednesday, June 3, 2009

Visualizing YouTube User Networks (

I recently ran across a very nice little tool for visualizing the connections between a YouTube user, his or her subscribers and his or her "friends". The tool, called YT Visualizer, is in beta but has worked very well since I started using it yesterday.

A trial version of the software is available for download from Loco Citato, the website of XMT Partners, a UK software design firm. (Note: I scanned the software for viruses and spyware/malware with two separate programs and found nothing but, as with all software you download, you do so at your own risk).

To get it to work (after you have downloaded and installed the software), all you have to do is enter the user name of someone who has uploaded a video to YouTube. For demonstration purposes, I used "almasri002".

Almasri002 has uploaded 323 videos to YouTube, many of which are videos that appear to support Islamic extremism. According to his "channel" on YouTube, he also has 488 subscribers and 567 "friends". Understanding more about this social network could conceivably be valuable in understanding the who, what, when, where, why and how of Islamic extremism on the internet but trying to do so manually, given the size of the network, would seem impossible.

YT Visualizer solves the problem by capturing and graphing all of this data into an easy to understand chart. Talking about it makes less sense that showing how it works so I built a little screencast of the tool in action (using one of my other favorite web-based tools, Screencast-o-matic) below:

I only let YT Visualizer run for a minute or so in this demo and, as a result, only managed to import 30 some odd people into the graph (YouTube gets fussy when you send too many requests too rapidly to their servers).

The full version of the software allows as many as 1000 entities to be imported into the graph (the demo version I have allows only 200). As the graph gets larger though, the software cleverly starts to fade out the less important nodes making the graph readable regardless of the number of nodes.

As you can also see in the demo, YT Visualizer pulls in some of the other data available from YouTube so that, when you mouse over the nodes you can see who they belong to, etc.

For the real social networking geeks, the best feature probably is the ability to download the data behind the visualization into a CSV file so that you can then upload it into other, more powerful programs like ORA, UCINET or Analyst's Notebook. This feature is disabled in the trial version so I was unable to test it out.

XMT has a number of other applications for visualizing data on the net but none of them are quite as polished or as powerful as YT Visualizer.

As cool as this is, I am pretty sure that it would be more useful for business professionals than national security types. I can see it providing some sort of contextual information for the national security intel analyst (though that context is pretty limited when you think about the constraints on the data set). I can also see it providing useful leads of the "I gotta start somewhere, so I might as well start here" variety.

Business/competitive intel analysts, however, might really be able to use this tool to identify influence hubs (at least on YouTube) of supporters and detractors of their products.

The potential is also there, of course, for speculative leaps far beyond the scope of the data but the possibilities inherent in the technology far outweigh the risks (for a careful researcher) in my mind.
Reblog this post [with Zemanta]

1 comment:

Kirk said...

This is pretty neat software. Thanks for sharing.