W3C home > Mailing lists > Public > public-vocabs@w3.org > January 2015

Re: Some interesting things that show up when using a reasoner to classify schema.org

From: <martin.hepp@ebusiness-unibw.org>
Date: Tue, 27 Jan 2015 22:50:35 +0100
Cc: W3C Web Schemas Task Force <public-vocabs@w3.org>
Message-Id: <E2620A87-B36E-4D37-B66C-981392AEF79B@ebusiness-unibw.org>
To: Thad Guidry <thadguidry@gmail.com>
Hi Thad:

I think we had a related conversation here on the list in July 2013:

   http://lists.w3.org/Archives/Public/public-vocabs/2013Jul/0145.html

;-)

In my opinion, we need the following:

1. A proper documentation of the schema.org meta-model and core modeling assumptions. That is hard and unrewarding work, admittedly. 
Only this will help us avoiding extensions and modifcations that break the meta-model. People will otherwise propose elements that duplicate or clash with existing patterns (e.g. QualitativeValue, datatypes, ...) or redefine existing elements.

2. A proper extension mechanism for both types and properties, and guidance on how and when to use the mechanism. This is why I tried hard to define the additionalType property proposal as a generic mechanism.

We are already seeing, and will see more, proposals that require 100+ new types. I do not think we have the resources to carefully review those for any potential clashes with other branches of the schema. IMO it will be better to clearly define and defend the borders of schema.org, while fostering the development and use of additional granularity in publicly endorsed external extensions.

3. Break with the notion of global property names and identifiers and bind their definition and identifier to the type they are used with.

Let's face it: schema.org is the biggest experiment of evolving a global schema in an open process. There have been schemas before that were used broadly, but most of the time determined by a single authority or small group. There have been academic proposals of how a large set of independent schemas can work in theory (but no proof so far). We simply do not know yet how this will work best, as the vocabulary, the user base, and the set of computational operations over the data grows. That's our mission ;-)


Best wishes / Mit freundlichen Grüßen

Martin Hepp

-------------------------------------------------------
martin hepp
e-business & web science research group
universitaet der bundeswehr muenchen

e-mail:  martin.hepp@unibw.de
phone:   +49-(0)89-6004-4217
fax:     +49-(0)89-6004-4620
www:     http://www.unibw.de/ebusiness/ (group)
         http://www.heppnetz.de/ (personal)
skype:   mfhepp 
twitter: mfhepp

Check out GoodRelations for E-Commerce on the Web of Linked Data!
=================================================================
* Project Main Page: http://purl.org/goodrelations/




On 27 Jan 2015, at 22:29, Thad Guidry <thadguidry@gmail.com> wrote:

> 
> 2. Strive for a self-contained, frame-based organisation, i.e. reducing the relevance of the type hierarchy, eventually up to a point where we (publicly) just have a flat bag of types and associated properties.
> 
> Uh, that breaks with the original goals of Schema.org somewhat ??... having a unified, agreed upon, schema (instead of everyone sorta doing their own thing and not having agreed upon types and properties)  I wonder what that flat bag of types would look like ?  Perhaps your thinking more that flat bag of types should be flipped around to Predicates themselves ?  Like something like Freebase's predicates, "abstraction of", "abstraction", "part_of", "parts" ... https://www.googleapis.com/freebase/v1/search?indent=true&help=predicates   but you still need basic types to hang a Predicate off of.
> 
> One of the "nice things to have" is that Schema.org PUBLICLY helps bring agreement on Types themselves.  I would not like to throw that out.
> 
> 
> That does not mean we abandon the hierarchy internally; it will remain useful for managing the vocabulary.
> 
> Then you'll need to be a little more clear...do we need a looser schema ? or only in some places ?  Remember the original goals of Schema.org and what we were trying to solve in the 1st place.  (Freebase created higher and higher kinded Types, but eventually you reach a plateau of sorts... I.E. We cannot just have Thing only in that flat bag. :)
> 
> 
> Currently, users and people who want to propose extensions must understand the official and inofficial parts of the meta-model and memorize the type hierarchy.
> 
> See Figure 4 in this paper:
> 
>     Possible Ontologies: How Reality Constrains the Development of Relevant Ontologies, in: IEEE Internet Computing, Vol. 11, No. 1, pp. 90-96, January-February 2007
> 
> A PDF is at http://www.heppnetz.de/files/IEEE-IC-PossibleOntologies-published.pdf
> 
> 
> Yeah, we know reality, Martin, and you and I agree a lot on that reality in the past :)   Hmm, perhaps I am just misinterpreting your global view of things for the future of Schema.org  ... can you elaborate more or blog about your flat blag of types idea ?
> 
Received on Tuesday, 27 January 2015 21:51:16 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 27 January 2015 21:51:17 UTC