SW CG Meeting

24 Feb 2010

See also: IRC log


Manu, Sandro, Scott, Chris, Jonathan Rees (guest)
Tom, Michael


IETF Link headers and known places

<ivan> http://tools.ietf.org/html/draft-nottingham-http-link-header-07

<ivan> http://tools.ietf.org/html/draft-nottingham-site-meta-05

jar: 2 RFC drafts, at least 1 or both are in last call
... these are related to linked data, and could provide a story

<ivan> scribenick: mscottm

<ivan> scribe: Scott

jar: OpenID application and WebFinger is related
... Let's start with CiteMeta and WebFinger
... Everybody has a user id. WebFinger - Associate metadata with e-mail address.

WebFinger - take hostname out of e-mail address to recreate finger functionality

<jar> google for webfinger... look google code site

<ivan> http://code.google.com/p/webfinger/

jar: WebFinger is from Yahoo
... You could perform an analogous exercise with RDF, i.e. with RDF, get FOAF files
... 1 piece of metadata that you're looking for is a public key.

CalDAV and Mozilla Autoconfig are using similar mechanisms to discover metadata

jar: Next about the 'link header', for providing typed links between resources, was in http 1.0, removed in http 1.1
... It's the essence of Semantic Web because you're linking resources (no RDF but the essence is there)

<jar> link: vs. <link>

jar: There is a certain response 'information ???'. How do you get triples for those resources?
... If the doc has HTML, then it may give you a linked element, that points you to further information.
... but it is hard/impractical to get the information from the doc itself
... If the doc is text/plain, how do you find out who the author is?
... It helps to have them separated (link and resource type).
... One possible solution: You could use a redirect service to provide the metadata.

<jar> <link> element may describe the representation (content), but not the resource per se

jar: So, http 'link header' is another route to get metadata and all that's needed to make it work is a simple agreement (W3C note or something).

<manu> http://payswarm.com/specs/web-api-discovery.html

<jar> cool

manu: [Q] We've got URL's that use a protocol, preferably using REST, maybe JSON. We're thinking about hanging an API Manifest off the web pages.
... Keywords that are associated with different URL's via the protocol

ivan: [Q] You mentioned that TAG was interested by this. What is the legal status now? Could it be official in 6 months?

jar: It's out there and waiting for recommendation.

ivan: [Q2] Let's say that I want to set up my website to return link headers. Can I do that simply with apache?

jar: I've done it in apache and it's a one liner.


jar: (IETF) There was some resistance to linked header but it's been generally accepted.

ivan: manu might want to pass that on to the RDF WG.

<jar> some on the TAG didn't want link: to compete with <link>

All: thanks Jonathan!


<ivan> Minutes of last meeting: http://www.w3.org/2010/01/27-swcg-minutes.html

Next meeting


manu: HTML 5 wanted to drop @ profile from head completely, as of last Friday that carried.
... Julian, Tantek, and myself were proposing to bring profile back to HTML 5 and allow it on every element. RDFa will benefit. Microdata would probably benefit.

ivan: Noting that GRDDL becomes unusable without the @ for HTML 5

ivan: Anything CG can/should do?

manu: We'll publish in about a month (March) so let's wait until everyone has something to consider.

scott: Shared Names is making progress on agreeing on a URI format for gene id's. Essential PURL infrastructure is almost there.

ivan: Maybe someone could talk about Shared Names at an upcoming SWCG meeting.

<sandro> sandro: No, RIF wont be at PR next meeting.


<ivan> rssagent, draft minutes

<danbri> oh my apologies; I forgot to send regrets :(