W3C home > Mailing lists > Public > public-vocabs@w3.org > June 2014

Re: Benifit of Schema.org over linked data or vice versa

From: Dan Brickley <danbri@google.com>
Date: Mon, 2 Jun 2014 14:40:55 +0100
Message-ID: <CAK-qy=6+-tchuPfB_Dn6sM+=9GpRaLLAzmXh5WSGKbE9O3oh=Q@mail.gmail.com>
To: Farhana Sarker <fs5g09@ecs.soton.ac.uk>
Cc: W3C Web Schemas Task Force <public-vocabs@w3.org>
On 31 May 2014 22:20, Farhana Sarker <fs5g09@ecs.soton.ac.uk> wrote:
> Hi,
> I am Farhana Sarker, student in the University of Southampton.
> I am keen to know the benefit of using schema.org over linked data technology in reusing or interoperability of data or vice versa.
> Could you please help me in this regards?

There is no rigid line separating "Linked Data" from other
RDF-oriented ways of sharing data. Schema.org can be seen as Linked
Data. But there are differences in emphasis across the community.

When TimBL originally wrote
http://www.w3.org/designissues/linkeddata.html it was largely a
response to the indirect linking model we'd been using in the FOAF
project. ("This linking system was very successful, forming a  growing
social network, and dominating, in 2006, the linked data available on
the web."). Tim was concerned that we (i.e. the FOAF people at the
time) were missing the opportunity to give re-usable real world
entities URL Web identifiers; instead FOAF descriptions tended to
indirectly identify entities, e.g. writing RDF that said "the Person
whose homepage is http://example.com/person123/", rather than giving a
formal URI for that Person (see
http://blog.foaf-project.org/2003/07/identifying-things-in-foaf/
http://blog.foaf-project.org/2003/07/missing-isnt-broken-data-validation-and-freedom-on-the-semantic-web/
for the FOAF approaches).  So anyway the Linked Data note advocated
for all entities to be given HTTP URLs that pointed at RDF
descriptions. This emphasis ended up becoming a central theme amongst
Linked Data advocates: the idea that real world entities should have
HTTP URLs that give you RDF data when you fetch them.

At that time, RDFa was also fairly new, although the effort was
underway, http://markbirbeck.com/blog/2004/04/01/xml-europe-2004-rdfxhtml-new-rdf-syntax/
... so Linked Data tended to be deployed using distinct
machine-oriented URLs rather than mixed-in with normal HTML content.

By the time schema.org was created (2011) it was clear that some of
the deployment habits considered "best practice" around Linked Data
were making life hard for publishers, especially those who knew
nothing about RDF. For example, the expectations that every mention of
an entity included an HTTP URL pointing at RDF data; that /-based URLs
redirected, that multiple formats were served using HTTP content
negotiation; that data used several independently managed RDF schemas,
etc. etc. ...

This 2008-style deployment of RDF as Linked Data has been pretty
successful in various professional settings - e.g cultural heritage,
archives, libraries
(http://www.w3.org/2005/Incubator/lld/XGR-lld-vocabdataset-20111025/),
but was too complex for many mainstream publishers, webmasters and
developers.

Schema.org makes different tradeoffs, but in the same design space.
Schema.org is an RDF vocabulary, even if we don't  shout about the RDF
side of it to publishers. The central emphasis is on ease of adoption
by publishers, even if this introduces more noise into the data. So
for example schema.org has been less emphatic about always
distinguishing URLs for things from URLs for pages about those things.
Those are still important distinctions to make, but we are still
somewhat in the early days of structured data going mainstream - it is
important to make things easy for publishers and introduce complexity
gradually.

The other "break with tradition" at schema.org is the focus on a
single integrated core vocabulary, rather than an overlapping
patchwork of independent schemas. Again this is motivated by ease of
publisher adoption, and by the importance of getting agreement amongst
high profile consumers. As with entity URLs it is best to consider
this a matter of emphasis rather than a rigid timeless principle.
There are lots of places in schema.org where external schemas could
usefully be combined with the schema.org vocabulary (some discussing
of this is in http://blog.schema.org/2012/05/schemaorg-markup-for-external-lists.html
).

And as Peter just mentioned, schema.org's vocabulary is defined using
RDFS a base but with some custom variations. In particular
schema.org's association of properties with types is very informal,
almost wiki-like. This is largely due to the fact that schema.org is
continually growing and evolving while published across millions of
sites, so we have tried to find ways of keeping things loose, so that
new type/property patterns can evolve without too much suprise.

To cut a long story short, schema.org is another effort in the larger
RDF family. It shares many classic Linked Data concerns but makes
various tradeoffs favouring broad adoption over data purity.

hope this helps,

Dan
Received on Monday, 2 June 2014 13:41:25 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:29:42 UTC