Jim Groom and I did a talk this week for ACCS on tagging and the semantic structured web. The upshot is looking into ways to use the structure and searching power of RDF to create new ways to make use of tagging. Here, it's looking at tags on blogs.

Below are a few examples of RSS feeds, plus a link to a SIMILE Exhibit, that demonstrate some of the possibilities. Instead of feeding out syndicated content, they're feeding out remixes of data coming in from RSS feeds (a lot like Yahoo Pipes, but when we get to the last example the differences will be clear). The RDF graph at the heart of it is based on a snapshot of the feeds from about 10 people, with some SIOC data added on top. It's all done with the help of RAP. I was getting some unexpected results here and there from RAP's feed parser (Magpie), so I sent the feeds through Dave Beckett's Triplr.

One important discovery was the myriad ways that info about tags gets put into RSS feeds. To get something like this working in a site-agnostic way, there will be lots of work to sort out how info about tags is being reported. That's complicated by the fact that WordPress categories are sometimes really tags, so I want to capture that info, too. In a nutshell, sometimes you get dc:subject to express a tag, sometimes <category term="(tag)" />, sometimes <category>(tag)<category>. UltimateTagWarrior users have feeds that use the rel attribute to signal a tag, so that needs to be scraped from the content.

First we have an RSS feed of the tags being used. The idea that I think makes something like this useful is the possibility that, exactly because of the subjectivity and ambiguity of folksonomic tags, they are useful for defining and understanding the community using a tag. Take, for example, the tag OER applied to lots of things in del.icio.us. Knowing what that is reflects and signals membership in a particular community.

If tags can be used to gain insight into a community, then a feed of the tags used by that community might help us to understand the community as it develops. The particular kind of community I have in mind is a university class. Watching the list of tags used might reveal how the class is developing from the students' perspective.

Here's a feed of the links made between this particular set of bloggers (yes, it's also reporting when people link to themselves--that'll be adjusted. Right now, the application is so pre-alpha it's cuneiform). If we believe Google that linking reflects something relevant (whatever that something is), then here we'd have a feed that might show what ideas are of interest to a class. Ever wanted to know more of what students are interested in talking about before class? Here ya go, and it works just by looking at the conversations, rather than the stilted and awkward ways that that info is often solicited.

Combining the two. Imagine looking back on a class and trying to figure out how to revise it. A good long list of the links along with the tags used on those links would say a lot about what topics generated the most interest and discussion.

Last, here's a link to the data in the RDF graph being reported out into a SIMILE Exhibit. Best thing to do is visit there and browse around the different views and the data. It's little more than a wire frame of what could be, but hopefully the idea gets across.