The project to produce a 'directory' (whatever that means when working with these kinds of sources) of UMWBlogs took an interesting turn a few weeks ago when I thought that getting at the connections between posts might be a more interesting starting point, and that the most natural way to expose those would be through a sidebar. So, I set to figuring out how to write a Firefox sidebar. It's been an adventure and education, but at this point things are coming along happily enough for me to post up the progress.

On the server side, I have ARC2 playing fairly nicely with SimplePie to create RDF data from the Atom feeds (see this post -- the dependence on feeds, rather than using a plugin, is because of the need to encompass other blogging platforms down the road ). The sidebar then queries the server, it does some SPARQL, and that's stuffed into the sidebar. Here's what that looks like so far:

Here, we've got the sidebar starting with details about the post itself, what it links to, what links into it, and the tags/categories used (not making any distinction, 'cuz across many blogs I don't think the distinction will hold). For each tag, there's also a space listing other posts using the tag.

The nifty thing about this is that, since it's just gathering data based on URL, links to things outside UMWblogs become a handy way to find posts inside UMW blogs. Here we see that someone has written a post about "New Media Studies" being a Wikipedia entry, so when I visit the Wikipedia entry I'm guided to the post.

This feature will make this extremely handy for discovering connections between courses. If our bloggers can develop the habit of including a link to Wikipedia for whatever keywords appear in their post (as I've done with Firefox and SPARQL), then here's an automatic way to discover who's writing about the same things. Of course, this will also let me play nicely with DBpedia sometime down the road!

From here, I still have lots of word to do to prettify the sidebar, especially to style the links so they look like links. I'm doing work now to improve the feed scraper, especially adding Dublin Core type info when I can (i.e., for images and audio embedded in the post). I'm also working on a script that will go out and look up titles for external links -- the first draft of it is currently churning away at the task.

(Oh...and I'll explain the name "Amiatinus" for the sidebar in an upcoming post)