'Social' Networking using OAI

I've recently been conducting some research on "collections understanding" among humanities faculty. While a number of studies have been conducted on the information-seeking behavior of humanties scholars, many of these ignore the role that knowing about collections plays in their research activities.

At a meeting yesterday I learned that in the Open Archives Initiative Protocol for Metdata Harvesting (OAI-PMH) has the ability for repositories to indicate that they have "friends." From the OAI-PMH Implementation Guidelines, "friends" are "a recommended schema allowing a repository to list confederate repositories as a means to support automatic discovery of repositories by harvesters." Here at UIUC Tom Habing is mapping the connections specified in repositories that say they have "friends" (See the UIUC OAI Experimental Repository for graphs in various formats).

While the Implementation Guidelines suggest this is really about discovery of other repositories, I'm now wondering what it means for institutions and collections to have "friends." Does it mean they have some institutional/administrative connection? Can collections be friendly with other collections that have similar or related materials (e.g. I'm a archival manuscript collection who has a friend that has a museum collection of physical objects from my creator...).

We can and do build some of these relationships by searching or clustering metadata with similar descriptions. But I can search Friendster and find other people with similar interests as mine, it doesn't necessarily mean they are my friends. If someone is my friend, I can also indicate what kind of friend they are (family, social friends, work friends, "in a relationship" or my favorite "it's complicated" - and we all know about those kinds of collections!). The XSD for the friends schema has a place for "friends:type" but I don't see any suggested values for types. The friends tag is implemented by a data provider, I wonder how these institutions are deciding who their friends are (do they ask permisssion?, do people just point to repositories they think should be friends?)

I hadn't really considered this when I was conducting my research, but from what I've learned, establishing these kinds of relationships could actually enable some of the "citation chaining" behavior that scholars engage in. And being able to specify what type of friend a collection/repository is could be useful.

I'd be interested in hearing about any research going on about social networking of organizations or individuals that might offer some suggestions. I'll post any interesting examples of the use of "OAI" friends.