foaf:interest property it’s possible for me to describe my interests (musical, technical, etc) in my FOAF profile.
The term has been specified so that it has a range of
foaf:Document, with the implication that the
foaf:topic of that
foaf:Document is what I’m interested in. Seem a bit convoluted? Maybe, but there are benefits…
Firstly, by using a URI rather than a simple literal value we have more flexibility. For example we can provide further information about the URI, ranging from simple things like the documents title, topic, etc through to describing it’s relationships to other documents on the web. We can also merge data using this URI allowing us to link people together, e.g. “show me everyone with an interest in the semantic web”.
The second benefit to using URIs is that there’s no need to maintain a controlled vocabulary of terms from which the user might select their areas of interest. No centralization for us.
However there is a downside to this approach, and I’ve previously grumbled along these lines. To usefully merge data about interests, we have to hope that users will consistently choose the same URIs. And that seems unlikely, as not all areas of interest have an obvious URI.
If I’m interested in XML, RDF, or SVG there are some obvious places to link to: either the specifications of those formats, or perhaps the relevant W3C Activity pages. But for more general topics (“Gardening”, “Cooking”) it’s hard to select a definitive URI.
Now we could just rely on users from a given community to crib from each others profiles and copy-and-paste the URIs that their friends use to describe their interests. Which is quite likely, and will work well enough, but still doesn’t seem ideal.
Also, if I’m building a user interface to allow authoring of FOAF data it’s friendlier to be able to provide suggestions as to what URIs people should use (actually it’d be friendlier to hide the fact that there’s a URI there at all, but that’s a separate issue). Difficult to do unless one uses a google “I’m feeling lucky” search and prompt the user for a couple of keywords. One might also use the Google search URI itself, but as the meaning of terms drift over time, this seems slightly fragile.
It dawned on the other day, that folksonomy might be the answer: i.e. rely on social classification to build us a vocabulary from which topics can be selected.
So I’m proposing to use del.icio.us as the source of
There are some other interesting angles to explore here. For example I can use the API to get items that a user has classified under a given tag. Each tag URI also exposes an RSS feed that includes metadata about the articles, and some basic details about each user. And then there’s Ben
experimenting with using del.icio.us tags to categorize his blog postings.