Weblogs.com provides a feed of updates blogs in real time. There are a number of methods to add geocoding to blog either via registration to places like geoURL, and/or via the inclusion of a variety of meta data in your blog home page (and even in your rss feed). Of course, there are a number of projects and APIs that map this data and allow for search of blogs (or rather, bloggers) near a location and so on.
As I had already put together some geographic visualization code as well as a geocoder and aggregator for news articles, I thought it would be fun to listen to the Weblogs.com feed, geocode the blogs in real time, and display them in DataSphere. To do this, rather than look for meta data, I decided to look for blogs that contain profile information on the home page. Actually, the demo I put together only looks for MSN Spaces blogs. Other blogging hosts provide profile information, but as a first step I went with Spaces (I certainly wanted to avoid Blogger - I doubt spammers give valid locations).
My expectation was that when I superimposed the area of sunlight onto the map, I would see some positive or negative correlation with the location of bloggers. That didn't quite happen. Here are a couple of screen shots for now illustrating recent posts around July 16th 2005 at 10:25 pm EDT:
As with news geolocation, the size of the pin indicates the number of documents (here, the number of MSN Spaces bloggers at that location). Note that my geocoder knows a lot about certain countries (including USA, UK, Canada, Iraq) and for others uses the capital as a proxy for all locations. This is simply to save me the time of long load up time of the full database during experimentation.
One thing to note is that China (at least on July 16th 2005 around 10:25 pm EDT) was producing the most content. Also note that the images show slightly different content as the data had been updating between shots.
Comments