I've come to believe that at least once a month, everyone's plan for the day should be as follows. 1. Check OUseful.info 2. Spend rest of day following up on what they read in the most recent post. The world would be a better place.

Today's installment involves this post on following the ripples of a conversation out through the blogosphere through links, trackbacks, comments, etc.

Since I've been collecting links data from as much of UMWBlogs as I can (see SemanticUMW, I thought I'd make some graphs of linkage info. The graph was produced by a quick SPARQL query against the data from UMWBlogs that looks for any resource such that two posts link to it. The resource could be anything that's linked to, but the posts are posts in UMWBlogs.

CONSTRUCT {
?p1 a sioc:Post .
?p2 a sioc:Post . 
?p3 a bibo:Webpage .
}

WHERE {
?p1 a sioc:Post ;
  sioc:links_to ?p3 . 
?p2 a sioc:Post ; 
  sioc:links_to ?p3 . 

FILTER (?p1 != ?p2)
}

(The actual query was a bit more complicated, with lots more filters to weed out unruly things like all those "Hello World!" posts, which all link back to the same UMWBlogs start and help page. But this is the basic idea.)

The graphs below were produced with IsaViz. They're a little hard to see here, but click any of them for a larger version.

Example 1 just shows a simple snippet of the graph. The resource off to the right (http://www.educause.edu/educatingthenetgen), highlighted in red, is the resource that several blog posts link to (the red arcs show the links).

Examples 2-4 show a more complicated portion of the graph, and we'll walk through some of the links. First, again we've got just one resource (http://blogs.ubc.ca/vince) that's linked to from two others.




There's actually a little more to it, though. The two things linking to that resource are FeedWordPress Widget: If you blog it, it is no dream and FeedWordPress Widget: If you blog it, it is no dream. The joke is in the URIs. UMWBlogs makes heavy use of syndication to republish content. So the apparent duplication is a product of one post being republished in another blog. This might be an interesting question, both technically and philosophically, to consider. For example, in chasing through what links to conversations, it throws in the idea that the same conversation could split into different directions from exactly the same content, but presented in different contexts. This would also show up in tracking retweets.

Let's chase just one direction and highlight those links in red.

And lets follow another step out, down toward the lower right-hand corner.

This is just a visual look at the current graph and chasing through links, bouncing from one to the next. The data is there to gather up at least part of the neighborhoods Tony writes about. I don't think SPARQL queries are set up for recursing progressively farther away from a starting resource -- a more clever SPARQLer might be able to tell us differently. But, with ARC you can write a SPARQLScript that could gather up the needed data pretty easily.

Of course, one would probably want to SPARQL out more information, like the tags on a post, and maybe data about the blog itself. The tags might be a nice filter, or even a way to query only for posts that use a particular set of tags.