Re: why is there no sha1 data in outbound/inbound nodes? - Julian Bond - com.googlegroups.social-graph-api

21 messages in com.googlegroups.social-graph-apiRe: why is there no sha1 data in outb...

From	Sent On	Attachments
ethaniel	20 Apr 2008 06:57
Manish	21 Apr 2008 09:53
ethaniel	21 Apr 2008 10:16
Brad Fitzpatrick	21 Apr 2008 10:25
ethaniel	22 Apr 2008 00:33
Brad Fitzpatrick	22 Apr 2008 10:40
ethaniel	22 Apr 2008 23:39
Brad Fitzpatrick	23 Apr 2008 12:02
Julian Bond	23 Apr 2008 12:41
Brad Fitzpatrick	23 Apr 2008 14:24
Julian Bond	24 Apr 2008 00:51
Danny Ayers	24 Apr 2008 02:21
Julian Bond	24 Apr 2008 02:39
Martin Atkins	24 Apr 2008 12:54
ethaniel	25 Apr 2008 13:36
Brad Fitzpatrick	25 Apr 2008 13:44
ethaniel	26 Apr 2008 01:39
Ben Laurie	26 Apr 2008 09:49
Dale Newfield	26 Apr 2008 10:06
Martin Atkins	27 Apr 2008 03:11
ethaniel	30 Apr 2008 11:03

Subject:	Re: why is there no sha1 data in outbound/inbound nodes?
From:	Julian Bond (juli...@voidstar.com)
Date:	04/24/2008 12:51:17 AM
List:	com.googlegroups.social-graph-api

Brad Fitzpatrick <brad...@google.com> Wed, 23 Apr 2008 14:24:35

See above. So how about OpenID as an IFP for matching? IFP? The Social Graph API already uses OpenID delegate tags as "me" links, but that's not what you're asking I don't think. What's IFP?

Sorry. IFP = Inverse Functional Property. See http://xmlns.com/foaf/spec/#term_mbox A property that has the quality that one person may have many of them, but any one value generally only applies to one person. Examples include email address, mbox_sha1sum, OpenID, Instant Message address, personal home page, Blog URL.

Take the use case where a Service uses the SG-API to find the contacts of a recently joined member and attempts to match them with members who have already joined up. You have to go through all the possible IFPs returned and try and match them. mbox_sha1sum works really well because you almost certainly hold a verified email address for every member and so can generate an mbox_sha1sum to match with the incoming data. You may not hold data for the other types and if you do they're probably not verified.

I don't expect FOAF mbox_sha1sum usage to increase much in the future.

Which is to say, you don't expect to see FOAF usage increase much. Should we be encouraging sites to produce more FOAF or is that a dead end (in your opinion).

Do you think we'll see more OpenIDs on the web, either as html microformat or RDF foaf?

Everybody's an OP nowadays. Tons of people have OpenID URLs. There's an increasing amount of microformats and RDF as well.

Yes, with Yahoo!, AOL and Blogger playing, there are indeed a ton of OpenIDs. But are there a ton of OpenIDs embedded in html? Does anyone, anywhere actually put their unique OpenID into a web page? Maybe there's enough OpenID delegation going into http headers and HEDA sections with real openID-Keys. I'd be interested to know. Of course this is getting confused by Yahoo!'s neat system where the end user doesn't need to know their real Yahoo openid, just "yahoo.com". Of course that's useless for matching.

Yes, people are putting OpenIDs into FOAF, right next to the mbox_sha1sum but outside that are they getting included anywhere else that the SG spiders can see?

If you're seeing increasing amounts of RDF, is that FOAF or other namespaces? See above.

I think Microformats are a good light weight solution but they have problems. The data ends up quite unstructured and really only lends itself to Key-URL pairs. If you need a structure like Owner+ServiceName+URL+URL like say: This person with this openid has a profile at Twitter with this URL and this RSS feed URL, it's awkward to do.