Re: Voting for CreativeWork property names

I'm all for simplicity.  A quick, totally unscientific, scan of schema.org Types showed up a few 'is' properties, almost zero 'has' properties and a few 'of's.  So if we can achieve what we need without 'has' and possibly without 'is' we will fit better with the schema style.

Jeff is right to highlight that 'instance' might not be meaningful outside of individual domain – that is one reason behind my personal choice of 'example / exampleOf'  The 'of' reflecting the style of the relationship between schema:Person and schema:Organization – member / memberOf.

To help think things through, a while back, I mocked up the data/relationships for a very small library on the Wiki <http://www.w3.org/community/schemabibex/wiki/Examples/mylib>, which might be worth revisiting.  It covers [what some may refer to as] the FRBR WEMI stack plus a form of super-work, movie, audioBook and a box-set.  In real life, some of the more conceptual entities may well be authoritative descriptions referenced from other sources, but for ease of understanding (and making the URIs clickable in a Wiki) they are all described in this single imaginary library.

Every physical or conceptual thing in this library is described as a schema:CreativeWork or sub-type thereof. The more concrete features being enabled by sub-typing (Book, Movie) for expressions and multi-typing (with Product, SomeProducts, IndividualProduct) for manifestation, item, etc.

The relationships between one entity and another (the conceptual, to the more concrete, and the inverse) being defined by a pair of generic properties on CreativeWork.  Hence instance and instanceOf.  The use of property names such as 'work' opens up the potential for confusion (especially in the library world) providing a 'work' link from what is effectively a FRBR:Item to a FRBR:Manifestation would have most cataloguers confused.  Whereas saying that an item is an 'instanceOf' a manifestation, I believe, is less confusing.

The argument that, in the eventual documentation, instance & instanceOf would be collated together and hence be more understandable to the more novice coder, also carries some weight.

Even though my personal preference was for example/exampleOf, I was happy to go with the majority vote.  From the many many emails and meting discussions on this, it is clear we will not total agreement – I think the best we can aspire towards is an equality in grudging acceptance ;-)



From: Jeff Mixter <jeffmixter@gmail.com<mailto:jeffmixter@gmail.com>>
Date: Friday, 17 May 2013 14:51
To: Ross Singer <rxs@talis.com<mailto:rxs@talis.com>>
Cc: Alf Eaton <eaton.alf@gmail.com<mailto:eaton.alf@gmail.com>>, Richard Wallis <richard.wallis@oclc.org<mailto:richard.wallis@oclc.org>>, "public-schemabibex@w3.org<mailto:public-schemabibex@w3.org>" <public-schemabibex@w3.org<mailto:public-schemabibex@w3.org>>
Subject: Re: Voting for CreativeWork property names

For what it is worth, I think that use of simple object properties (such as has and is) work quite well, especially if one intends to use them outside the scope of library material.  I have been working on a developing a linked data model for VRA (a XML based schema used to catalog visual/cultural items) and having the ability to use simple object properties such as has and is would be hugely beneficial and help eliminate the need to create tons of custom object properties (such as hasPrint, hasPhotograph etc). The isInstance/hasInstance, while useful for the library community, might not be widely used by people in other fields, even related fields such museums or archives.

Jeff Mixter
440-773-9079


On Fri, May 17, 2013 at 8:39 AM, Ross Singer <rxs@talis.com<mailto:rxs@talis.com>> wrote:
On May 16, 2013, at 6:45 PM, Alf Eaton <eaton.alf@gmail.com<mailto:eaton.alf@gmail.com>> wrote:

Sorry, ignore my use of "expression" if you like; I just meant that simple names for object properties, that work as either "has x" or "is x of" when read in opposite directions, can make things easier to work with. It could just have easily have been this:

<http://www.worldcat.org/oclc/38264520> <http://proposed-schema.org/work> <http://exampleworks.org/work/12345>.
<http://exampleworks.org/work/12345> <http://proposed-schema.org/instance> <http://www.worldcat.org/oclc/38264520>.

In other words, maybe there's a better term than "instanceOf" which describes the relationship of <http://exampleworks.org/work/12345> to <http://www.worldcat.org/oclc/38264520>. What is <http://exampleworks.org/work/12345>, if not a "Work"?

I'm all for this suggestion, assuming that the object will always be a Work.  I got the impression that this was kind of taking the same approach as commonThing [1], where the subjects and objects can be of ambiguous types, but maybe I'm conflating some unrelated threads here.

-Ross.

1. http://dilettantes.code4lib.org/blog/2011/11/why-do-we-obsess-over-frbr-entities/


Alf

On 16 May 2013 23:12, Wallis,Richard <Richard.Wallis@oclc.org<mailto:Richard.Wallis@oclc.org>> wrote:
Hi Alf,

The approach proposed was shaped by several factors including:

  *   CreativeWork describes "The most generic kind of creative work, including books, movies, photographs, software programs, etc."
  *   It is the super type for many specific types such as Map, Painting, Movie, Book, Sculpture, etc.
  *   Schema.org<http://Schema.org> is a generic vocabulary with a broad consumer community therefore domain specific terms should be avoided if possible
  *   We have specific guidance that Schema.org<http://Schema.org> will never implement FRBR

On that last point, your suggestion is leaning in a FRBR direction. I can hear the follow on "we need  manifestation & item properties"  already.

Expression also has certain library-ish connotations

CreativeWork->work I would suggest is a little confusing as to

The hasInstance / instanceOf pair were proposed as generic and directional properties.

Following the vote about the need of 'is', I have some sympathy with the suggestion of dropping the 'has' making it 'instance / instanceOf'

~Richard

From: Alf Eaton <eaton.alf@gmail.com<mailto:eaton.alf@gmail.com><mailto:eaton.alf@gmail.com<mailto:eaton.alf@gmail.com>>>
Date: Thursday, 16 May 2013 18:30
To: "public-schemabibex@w3.org<mailto:public-schemabibex@w3.org><mailto:public-schemabibex@w3.org<mailto:public-schemabibex@w3.org>>" <public-schemabibex@w3.org<mailto:public-schemabibex@w3.org><mailto:public-schemabibex@w3.org<mailto:public-schemabibex@w3.org>>>
Subject: Re: Voting for CreativeWork property names
Resent-From: <public-schemabibex@w3.org<mailto:public-schemabibex@w3.org><mailto:public-schemabibex@w3.org<mailto:public-schemabibex@w3.org>>>
Resent-Date: Thursday, 16 May 2013 18:30

On 16 May 2013 17:55, Wallis,Richard <Richard.Wallis@oclc.org<mailto:Richard.Wallis@oclc.org><mailto:Richard.Wallis@oclc.org<mailto:Richard.Wallis@oclc.org>>> wrote:

> I have reflected these choices in the proposal page <http://www.w3.org/community/schemabibex/wiki/CreativeWork_Relationships>
> If people are happy with the proposal, I suggest that we should add some html examples to the turtle and then submit to public-vocabs.


Would it be more straightforward to make is/has/of implicit and just use simple property names that read well in both directions? Like this, for example:

<http://www.worldcat.org/oclc/38264520> <http://proposed-schema.org/work> <http://exampleworks.org/work/12345>.
<http://exampleworks.org/work/12345> <http://proposed-schema.org/expression> <http://www.worldcat.org/oclc/38264520>.

Perhaps this has been proposed and rejected already?

Alf






--
Jeff Mixter
jeffmixter@gmail.com<mailto:jeffmixter@gmail.com>
440-773-9079

Received on Friday, 17 May 2013 15:15:47 UTC