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

Re: Non-schema.org ontologies

From: Dan Brickley <danbri@google.com>
Date: Thu, 1 Sep 2016 15:42:30 +0100
Message-ID: <CAK-qy=4WqVFbiokkro24M9zB0SiCSC3FsRbOx8EQ9v-Qs72k_w@mail.gmail.com>
To: Timothy Holborn <timothy.holborn@gmail.com>
Cc: "schema.org Mailing List" <public-schemaorg@w3.org>
On 31 August 2016 at 16:14, Timothy Holborn <timothy.holborn@gmail.com>
wrote:
>
> If an ontological functions wasn't described by schema.org ie: amenities:
toilets or as may be similarly described, what other ontologies are scraped
by various search engines / A.I. Related services??


Is this just an example, or the main concern here?

In the CSV on the Web W3C WG I co-chaired with Jeni Tennison recently,
public toilets came up as an example often enough in conversation (f2f it
seems, I find nothing much in
https://lists.w3.org/Archives/Public/public-csv-wg/ ) as a case where the
CSVW facility for turning tablular data records into triples/graphs would
be a great usage example. CSVW includes machinery for expressing a template
based mapping from table rows into RDF. And the rhetorical point within the
WG was that there were two interestingly different discovery usecases: some
people might be searching for amenity-related datasets; others might be
looking for a toilet/bathroom to use.

At that point, and now, schema.org didn't actually have anything like a
PublicToilet type.

The most natural attachment point for this is
http://schema.org/CivicStructure, which also has
http://schema.org/openingHours and
http://schema.org/openingHoursSpecification

It is worth noting in passing also that we also have a discussion at
https://github.com/schemaorg/schemaorg/issues/254 towards better
description of the physical accessibility of a place. Also tha the latest
schema.org release (the recent version 3.1) includes a general construction
http://schema.org/amenityFeature +
http://schema.org/LocationFeatureSpecification to capture a long tail of
location-related features that are not explicitly anticipated by schema.org.

It can be easy for admiral goals (handling a longer tail of geo/spatial
kinds of place e.g. by using external lists such as from Wikidata) stand in
the way of simple easy improvements, such as adding PublicToilet directly.
I'd support doing so if the rough consensus here is that this would be
useful progress. We also recently have the "pending.schema.org" area of the
site where proposals can be aired and refined. Should we go ahead and add
PublicToilet there?

Backing up from these detail - and looking at the larger question, "what
other ontologies are scraped by various search engines / A.I. Related
services", and reading "parsed" for scraped, I am not aware of a comparable
case where search engines (or AI-oriented services) have come together
around a set of schemas/ontologies in the way they have for schema.org.
Probably in various settings you'll find some use of Dublin Core, of Open
Graph Protocol (OGP), some exploitation of Freebase's (now discontinued)
RDF dumps, and of Wikipedia data (perhaps increasingly via Wikidata). All
of these are expressible within the same RDF-oriented data model that is
used at schema.org <http://schema.org/docs/datamodel.html>. If all you want
to do is make linked data or RDF triples with a type for "public toilet",
DBPedia is already available - http://live.dbpedia.org/page/Public_toilet.

So a natural question at this point is whether progressing "public toilet"
on its own makes sense, versus taking somewhat longer but fleshing out a
larger set of things like "water fountain" to live alongside it...

Dan
Received on Thursday, 1 September 2016 14:43:00 UTC

This archive was generated by hypermail 2.3.1 : Thursday, 1 September 2016 14:43:00 UTC