I spent some more time today reading through the details of XFN (XHTML Friends Network) and XMDP (XHTML Meta Data Profiles), and, in the process, learned several things about HTML and CSS that I hadn't previously known. Plus, I definitely got a deeper sense of XFN and XMDP and think they are quite cool!

So, XMDP is a simple format (it is just a subset of XHTML) for defining HTML 4 Meta data profiles, which I had not previously known about. Basically, these profiles indicate metadata properties that can be added to HTML documents, and XMDP is a simple way to list and define the properties that are indicated.

This looks like a useful way to simply add some semantic meat to HTML documents. I might describe this as a simpler way to accomplishing some semantic tagging along the lines of XML + namespaces (and, with XMDP being something like XML Schema). Of course, being simpler, it does a lot less than XML + namespaces + XML Schema. But, as some might say, maybe we need some more practice doing a lot less in this area before we spend so much time trying design comprehensive ways of doing more!

So, XFN is a profile (defined by this XMDP document) for adding interpersonal relationship metadata to HTML links. As the XFN examples show, this is super simple: just a rel attribute on your A tags with a list of properties indicated in the profile. These are personal relationship attributes like: met, friend, acquaintance, etc.

What I also wasn't aware of was how many ways attributes can be used as selectors in CSS 2 style sheets. Tantek Çelik, one of the XFN creators, is a major CSS guy, and so he shows some interesting CSS examples for XFN (scroll down to the section titled "Relationship Styling").

This ability to immediately connect the XFN metadata to CSS style changes is very significant. I think many people associated with semantic web concepts maybe underestimate how important the relationship is between presentation styles and semantics. Yes, they should be declared separately (i.e., semantics separate from style) for utmost flexibility, but at whatever point humans work with the information / data, semantic subtly must be convenient to translate to presentability or else people generally fail to recognize much value in all of the meaning.

Finally, I was thinking about XFN parsing and generation by iCites, and thought of one example of how the iCite net and XFN might be good compliments:

In the iCite net, there will be a definitive way to indicate that a URL is the URL for a person, as possibly different from their blog or website home page, etc. (i.e., especially for people with more than one website or blog or who post on multiple sites).

In the iCite net context, the relationship between any webpage and the creator person URL is also definitive—so, if an XFN link is, for example, in a blog role and linked to another blog home page, and one or both blog URLs are different than the person URLs, the iCite net could track the XFN relationships back to the person URLs.

Something like:

person <-> blog <-xfn rel-> blog <-> person

could be seen in the iCite net as:

person <-xfn rel-> person

Also, the person-identity iCites should end-up representing a superset of information that can be used to generate both things like FOAF documents and HTML documents with XFN links, which would be a beautiful thing.