Re: Finalising bib.schema.org 1.0 proposal

> On 22 Apr 2015, at 16:39, Wallis,Richard <Richard.Wallis@oclc.org> wrote:
> 
> To address a few points that have come up in this thread:
> 
> Scope of an extension domain / splitting into several extensions
> 
> There are three aspects to this.
> Does a proposed term sit well in an extension with a particular name
> Is there a community ready, and most importantly willing, to focus on and maintain a particular subdomain
> Does the term in question satisfy 'a need to describe’ in the sponsoring [of an extension subdomain] community.
> 
> Taking comics as an example.  If you have a fairly narrow definition of what bibliographic means, and only address point 1.  it leads towards a conclusion that comics.schema.org <http://comics.schema.org/> might be a good idea.  However, taking all three aspects into account and recognising that much of what is in comics is an extension of bib properties anyway — as Dan put it “There's a ton of overlap in the bibliographic domain and the comics domain” — I think pragmatically that there is not much problem with it being part of a bib extension.
> 

I think the argument for having a separate Comics extension is related to (2) - that there might be a more specialist community ready and willing to focus on it.

However I’m happy to see it as part of the bibliographic extension - I definitely wouldn’t argue that Comics are not ‘bibliographic’ in nature and clearly the overlap between comics and other types of bibliographic material is substantial.

> 
> Is a Globe, etc. a bibliographic item or not
> Following on from the above, especially with reference to point 3.
> 
> If a suitable term (property or type) does not already exist in core Schema, or an extension to Schema; and there is no obvious group who could be lobbied to include it in their domain; and members of the domain(s) our community represent/support have a need to describe that property or type of thing; then it is within scope of our group to propose it.  In doing so there should always be a view to any term being used elsewhere across the full breadth of Schema.org <http://schema.org/> and its extensions.
> 
> Grounding it in the Globe example — There is not already a Globe type, no one else is lightly to define one, members of our community (libraries in this case) do need to describe globes in their collections.

I’m not completely convinced by this. I agree that the community should be free to define the necessary attributes, but I think there *is* something in a name. What would be the cost of this community supporting multiple extensions?

> 
> Do you need properties to justify the need for a subtype
> There are several Types in Schema.org <http://schema.org/> that do not have additional properties compared with their super-type.  This is an establish pattern for two reasons.  Either the subtype has no other obvious properties, being a different type is sufficient to differentiate it from the super-type; or the definition is a foundation definition for others to elaborate upon in later releases.
> 

Agreed.
The issue is that for some things a ‘type by property’ may be sufficient and in others a ‘type as type’ is needed. It feels like these decisions have been very ad hoc in Schema.org and it freely mixes these approaches. I think *one* of the questions you can ask to help define the right approach for any particular type of thing is to ask ‘does it have a set of specific properties’ - because if the answer is ‘yes’ you probably need a new type.

For me this is a very quick litmus test - the lack of specific properties suggests to me that there needs some further thought/justification before  going ahead. But it is just an indicator, and I understand it isn’t the only factor and that it is OK to have new types without specific properties.


> Agent
> The proposal for Agent effectively creates a new super-type for Person, Organisation, and BroadcastService.  
> 
> The initial reasoning being to satisfy the real cases where the creator etc. is unknown as to its type.  In an ideal world the describer of a resource would know if the creator was a Person or an Organization, but as we know in real life that is not the case.  I expect that several other types will become candidates for Agent subtypes in the future, SoftwareApplication for example.
> 
> Secondly, it provides a way of rationalising documentation of ranges of some properties to include Agent instead of having to list all the possible types.

So this means the description and the implementation differ (as the description mentions Software explicitly)
What does Agent add that Thing doesn’t already offer?

> 
> 
> This is getting a bit long so I’ll stop for now.
> 
And me!

Owen