Re: Question on proposed sports vocab extensions for schema.org

On 5 March 2012 15:57, Paul Wilton <paul.wilton@ontoba.com> wrote:
> Hi
> I have been working recently at the Press Association with the Sport
> ontology, modelling the Olympics , Football and Horse Racing. I have
> been following the proposed sports vocabulary extensions for
> Schema.org, and have a question around these constructs :
>
> Thing > Person > SportsAthlete
> Thing > Person > SportsAthlete/Baseball
> Thing > Person > SportsAthlete/Football
>
> I guess I am not 100% comfortable with the idea of sub-classing a
> Person as an Athlete - I dont really see an Athlete as a more
> specialised form of a Person, but rather than a Role or Vocation that
> a Person takes in life.

I understand the discomfort, and it is natural to want to model roles
explicitly in some situations. But there's nothing intrinsically wrong
with creating named subclasses of Person, when it corresponds to some
reasonably clear definition that picks out a useful subclass of
'person'.

The schema.org design tends towards designs that correspond to simpler
markup (e.g. less nesting, fewer subelements etc.), even at the cost
sometimes of elegance in the underlying model. Whether that's the case
here I'm not sure without working through some detailed examples.

> Just because we tend to think of many roles/vocations as being people
> oriented (Doctor, Engineer), in many cases they are not. For example
> in this case ESPN list "Secreteriat" , a race horse, in their top 50
> Atheletes of the century (number 36)  :
>
> http://espn.go.com/sportscentury/athletes.html
>
> Thus, if ESPN marked up their list against this vocabulary, we would
> be inferring that Secreteriat is a Person.

...which would be a fairly straightforward error, agreed. We don't
have a class for animals or agents yet. I don't see any microdata
markup in those pages. It would be good to know from ESPN whether
distinguishing human from other 'athletes' is relatively easy for them
(I'd guess that it wouldn't be too hard).

> So I am keen to understand the reasoning behind this approach , or of
> this could possibly be changed such that Roles (temporal) /or
> Vocations (possibly not temporal)  are separate Things that are
> applied via some other relationship to a Person (or a Horse or a Robot
> for that matter)

Bear in mind that each level of indirection / richness we introduce in
the modeling will need to show up somehow in HTML markup, and this is
a very error-prone business. Let's try to get complete examples of
both styles into the wiki. Does anyone have a page handy that can be
used as a basis?

cheers,

Dan

Received on Thursday, 8 March 2012 22:41:08 UTC