W3C home > Mailing lists > Public > public-schemaorg@w3.org > April 2016

Re: Review schema for books and short stories.

From: Dan Brickley <danbri@google.com>
Date: Sun, 10 Apr 2016 22:26:33 +0100
Message-ID: <CAK-qy=5Ft544vP2ac0Z_hcEB6UjBTgQfairJMoptKyMxiidL+w@mail.gmail.com>
To: Alex McKee <alex@agmckee.co.uk>
Cc: "schema.org Mailing List" <public-schemaorg@w3.org>
On 8 April 2016 at 11:26, Alex McKee <alex@agmckee.co.uk> wrote:

> Ali,
>
> Bear in mind that older books often do not have an ISBN. Anything
> published before the mid 1960s will have no ISBN. So I strongly oppose
> making ISBN required.
>

I agree. Most importantly, even if all books in the universe have ISBNs,
that doesn't mean that everyone with useful book data always has that
information. And more generally let me emphasize something important:
schema.org itself does not make anything required, all (*) it does is
provide definitions for terms you can use in structured data.

The best practical analogy for understanding the role of schema.org is the
dictionary. It would be easy to take this too literally and get
side-tracked into debating meaning and machine understanding and the nature
of language and so on, but the main point is just this: a dictionary gives
short summaries of how terms are used and combined, but it does not tell
you what to say. A dictionary does not care about whether you have said too
much, or too little. That is not its role, and it would not be good at that
role. A dictionary can help sometimes in understanding when you have said
something self-contradictory, but it is generally of limited use for
understanding whether you have said something intelligible or useful or
coherent. Schema.org is similar: it gives some definitions that can be used
for saying things via structured data, but it has no notion of there being
required pieces of information.

The closest we get are the numerous examples posted on the site, which are
provided for several reasons. 1.) as a gentle nudge towards establishing
common patterns for combining terms 2.) as an alternate form of
documentation of the meaning of terms and how they can be combined to
represent sometimes complex situations 3.) as a practical way to help teach
the underlying syntaxes that are used by schema.org (i.e. Microdata, RDFa,
JSON-LD).

It is quite reasonable, on top of all this, for specific tools, services
and products to have their own stricter data requirements. So e.g. some
bibliographic systems might be unable to do something useful unless you
give them an ISBN, and might at that level have a requirement for ISBNs in
the data. But lacking the ISBN is not a fault with the data, only with its
suitability to some practical task. As far as Google's Structured Data
Testing Tool goes, its validation 'errors' are very much in this style: it
often doesn't mean that there is anything wrong with your data as such,
just that it lacks something needed for a particular product or feature.

cheers,

Dan

p.s. I wrote about this a long time ago in a different setting,
https://web.archive.org/web/20031003223643/http://rdfweb.org/mt/foaflog/archives/000047.html
... and more recently for https://www.w3.org/2012/12/rdf-val/ in a position
paper with Shawn Simister,
https://www.w3.org/2001/sw/wiki/images/0/00/SimpleApplication-SpecificConstraintsforRDFModels.pdf
- which is why Holger's exploration at
https://lists.w3.org/Archives/Public/public-schemaorg/2016Mar/0077.html is
interesting. It might be possible to use W3C SHACL as a way to capture the
requirements and patterns of sites and services that *consume* schema.org
in a machine processable way.

p.p.s. (*) I slightly over-stated the case. the schema.org Actions
mechanism http://schema.org/Action http://schema.org/docs/actions.html does
actually touch on these issues -- it is the only place where schema.org
crosses over from describing the world to indicating how to actually effect
changes in the world. As such we defined some lightweight mechanisms
inspired by HTML form validation that are used to indicate which pieces of
information are needed to request that a potentialAction actually happen.
Something like SHACL could also be an interesting technology to apply to
this usecase.

On 05/04/2016 04:26, Ali Kayn wrote:
>
> I do reviews.  For a collection I still use 'author' for the editor, as a
> broader interpretation of the word.
>
> I absolutely support requiring an ISBN for books.  Books and Magazines
> (ISSN) are specific object that are required to be archived  by state and
> national libraries.
>
> SameAs can be a bit harder to track down.  Sometimes books and stories are
> listed on author sites, sometimes they are listed on Amazon or Wikipedia.
>
> Reviewing short stories within collections looks like something not yet
> considered.  If you do stories within a collection or a magazine, can you
> use itemlist?
>
>
>
>
> ALi Kayn
> Editor and Publisher
> Festivale Online Magazine
>
> http://festivale.info
> http://reel-life.info
> https://facebook.com/FestivaleOnline
> https://www.linkedin.com/in/ali-kayn-9a608b45
> twitter:  @ali_festivale
>
> GPO Box 1510
> MELBOURNE GPO  VIC  3001
> Australia
>
>
>
>

picture
(image/jpeg attachment: 01-part)

Received on Sunday, 10 April 2016 21:27:03 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:12:24 UTC