2 messages in com.googlegroups.social-graph-apiRe: Special case for rel="me" for ide...
FromSent OnAttachments
Martin Atkins07 Jul 2008 11:13 
Brad Fitzpatrick07 Jul 2008 11:26 
Subject:Re: Special case for rel="me" for identity consolidation
From:Brad Fitzpatrick (brad@google.com)
Date:07/07/2008 11:26:01 AM
List:com.googlegroups.social-graph-api

No, that makes sense and could definitely be useful. I'd have to think how to compute that efficiently, though.... there's some tricky details. I'll file a low-priority feature request bug.

On Mon, Jul 7, 2008 at 11:14 AM, Martin Atkins <ma@degeneration.co.uk> wrote:

This page has a section 'Identity consolidation with the XFN rel="me" value': <http://www.gmpg.org/xfn/and/>

The following paragraph caught my eye:

Since such sites typically do not allow editing the "rel" attribute of a user's pages, this usage introduces a special case for handling the "me" value. For sites that allow you to enter a URL or home page or blog URL as part of a person's profile page, the "me" relationship can be confirmed with a bidrectional link, where the link to those domains has a rel="me" attribute, even though the link from those centralized social networking domains does not have a rel="me" attribute.

Does SGAPI currently do anything about the above? If not, it'd be useful if information about these un-relled reciprocal links could be included in the list of edges somehow so that applications can make the above assumption where appopriate.

The best representation I can come up with off the top of my head is a "nodes_referenced_by" entry with an empty "types" list. This would obviously only be included when the referee has an explicit rel="me" link to the referent, or else SGAPI would just become a big graph of all of the links on the web!

Does this sound feasible, or am I asking for the moon on a stick?