W3C home > Mailing lists > Public > public-lod@w3.org > July 2013

Re: Schema.org Autodiscovery?

From: Dan Brickley <danbri@danbri.org>
Date: Fri, 12 Jul 2013 23:29:36 +0100
Message-ID: <CAFfrAFoRPmQq_wFRmcj=8pB8=sEXmuLfcRqOQM5n2OX6BHyWzw@mail.gmail.com>
To: David Wood <prototypo@me.com>
Cc: public-lod org <public-lod@w3.org>
On 12 July 2013 21:08, David Wood <prototypo@me.com> wrote:
> Hi all,
>
> Does anyone (Dan?) know whether the schema.org folks do or plan to support Linked Data referenced via <link rel="meta"> tags?  The idea would be very similar to FOAF Autodiscovery [1].
>
> I would think it would be quicker for the search engines to find and parse a dereferenceable Turtle file than to parse RDFa from HTML, but perhaps I'm mistaken.

Generally schema.org has focussed on schemas, with a sideline in
saying encouraging things about RDFa and JSON-LD. There has been a
central emphasis on in-page markup mainly because of a suspicion of
hidden markup. Search engines are on the receiving end of a lot of
mischievous markup, and having structured data be part of human facing
markup is a good strategy for keeping it relatively honest. That said
JSON-LD also has a role to play (e.g. see
https://developers.google.com/gmail/schemas/reference/formats/json-ld
) and there were some schema.org-related conversations around
http://www.w3.org/TR/json-ld/#embedding-json-ld-in-html-documents ...
which can be considered a form of discovery.

Regarding speed/details, what little I know is only relevant for
Google. HTML pages are crawled anyway, for ordinary search purposes. I
haven't heard anyone argue that Turtle is particularly needed, but in
general I think you'd find more enthusiasm for a link-rel-based
discovery mechanism compared to HTTP content negotiation or HTTP 303
redirects. JSON-LD does cover some of the same use cases as Turtle.
But there are places where Turtle is useful, e.g.
https://developers.google.com/freebase/data#data-format ... I'm not
sure it's strengths are in the mass market  'everyday webmaster'
world, though. And the fact that Turtle and SPARQL are so close makes
Turtle a great language for documenting and displaying triples in a
lossless and reasonably human friendly way.

On the original question, it might be worth documenting a few other
deployment routes, e.g. relationship to sitemaps, rss/atom feeds etc.

Dan
Received on Friday, 12 July 2013 22:30:03 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 15:16:38 UTC