As an experiment, I just wrote a small crawler to parse the links and mine some XFN data. It was fun, and this post is an analysis of the collected data.

I did end up with thousands of links and relations, but for the analysis of this post, I am going to ignore those collected from “social”sites (twitter, digg etc), because of two reasons:
First, most people aren’t serious about whom they add as “friends”. For example, over 90% of my “friends” in Facebook, are only Facebook friends.

Second, many of these sites either add default values to the rel tags and/or don’t allow the users to edit them. For example, Gaia adds “contact” to the rel tag of my “friends” list, by default.

It is difficult to get accurate idea of one’s relations list, because of the above two reasons. So, for this post, I excluded the “social” sites data, even though it meant excluding over 90% of the collected data. This is the resultant graph

and the raw data (decreasing order of usage of tags):

The data set is pretty small, just a little over 2300 relations. But this is enough to give us a general idea of which XFN tags are used heavily and which aren’t.

From the table, we can see that just four relations - friend, met, me and colleague account for the bulk of the XFN tags used, nearly 80%. The current XFN spec is a balanced one, but nearly 14 out of the 18 rel tags, aren’t used much at all. May be its time to simplify the spec?

I’m going to collect links here in this post. Please mention your favorite links in the comments. I’ll keep updating this post with the links as and when I find/get them.

Tamil Nation - Nice site, has a few of Kalki’s novels

Chennai Library - Has works of Bharathiyar, Kalki and a few others

Noolaham - Very nice site, contains hundreds of PDFs to old Srilankan Tamil magazines and books. Amazing.

Facebook groups:

Noolaham