Re: AdditionalType was: audiobook options in objects

When I saw Dan Brickley talk about Schema.org a little while back (watch it at http://www.youtube.com/watch?v=_-6mhdjE1XE) the thing that struck me is how incredibly pragmatic the approach was - it was about 'how do people currently represent this on the web' not 'how best to represent this'. I keep having to remind myself about this when I think about making proposals.

With this in mind I've followed Karen's example and started to look at how Audiobooks are described on the web - I'm keen that whatever markup we propose is going to support these examples. I've started collecting examples and added them to the wiki. I did start to work out how these might be supported by some of the proposals with no, or only small, changes to the existing HTML markup - but haven't had time to complete this yet.

It would be good to get some links to existing library specific displays of audiobooks as well - don't have any of these yet, so please add to the wiki if you have some.

I guess that I'm trying to get into what I think is the schema.org mindset rather than a more general modelling mindset and ground proposals in real world existing html markup. I'm keen that we ground proposals in real world stuff, and think this is a way of ensuring this is what we do. To my mind this is a strength of discussing specifics like Audiobooks over the more abstract content vs carrier discussion - if we do this for some key types that exemplify content vs carrier, we may find a set of consistent approaches that all work in the same way, or we may find that we need different approaches in different areas - but we shouldn't worry either way.

I think we'll stand a better chance getting three proposals for  "Audiobook", "Radio Play" and "TV Show recording"  to be added than a single, more abstract, how to do content vs carrier proposal.

I'd be interested in knowing if this strikes a chord with others

Owen

Owen Stephens
Owen Stephens Consulting
Web: http://www.ostephens.com
Email: owen@ostephens.com
Telephone: 0121 288 6936

On 13 Feb 2013, at 14:03, Karen Coyle <kcoyle@KCOYLE.NET> wrote:

> Richard, I don't think that we can declare that each bibliographic description describes a single, uncomplex type. To begin with, there is that library bugaboo "kit" in which the item in question is simultaneously multiple types:
>   a kit with multiple parts, each of which is a different thing (a puppet, a book, some crayons)
> 
> There is also:
>   a book with an included CD
> 
> There are also many libraries that do not create separate records for the hard copy and digital:
>   record for a book with an additional link to the online copy
> 
> And almost none create separate records for hardcopy and paperbacks.
> 
> The upshot is that we will need to handle multiple types in a single description. These are also an "AND" relationships, at least in relation to the bibliographic data. How would this be done?
> 
> [And in another thread, as I say, I do not consider a "CD" to be a further typing of a creative work, since I would not say that a "CD" is a type of musical work.]
> 
> kc
> 
> 
> On 2/13/13 6:57 AM, Richard Wallis wrote:
>> Hi All,
>> 
>> I’ve pulled this out of the audiobook thread as I think it is generally
>> applicable to several areas of our discussions.
>> 
>> Karen’s points below highlight several points relevant to this, which I
>> will try to clarify.
>> 
>> This emerged from the audiobook thread as audio book is a good example
>> of something in our domain of multiple types – a creative work, possibly
>> a book, with a file format (WMA, MP3, etc), and a physical form (CD,
>> cassette tape, etc.).  That thread has moved on and we proposing a new
>> sub-type of CreativeWork – AudioBook, which I agree with.  For the
>> purposes of examples in this email am presuming that proposal has been
>> accepted.
>> 
>> Starting with Karen’s second question:
>> 
>>    /Second, it still isn't clear, however, if you have multiple
>>    associatedMedia fields, e.g. A, B and C, whether that means that you
>>    have that CW in three different media, or if you have the CW in a
>>    single medium that is defined as A+B+C.
>>    /
>> 
>> 
>> She is referencing multiple instances of a property, however I believe
>> it is the same question for multiple types.
>> 
>> It is an AND relationship.
>> 
>> The turtle syntax is really helpful for envisioning multiple types:
>>     <http://example.com/1234>
>>         a schema:Audiobook, pto:Windows_Media_Audio, pto:Compact_Disk;
>> 
>> Which can be unpacked as:
>>     <http://example.com/1234>
>>         a schema:Audiobook;
>>         a pto:Windows_Media_Audio;
>>         a pto:Compact_Disk;
>> 
>> Which can be read as:
>>     <http://example.com/1234> is the identifier for a thing which is
>>         a Audiobook and,
>>         a Windows_Media_Audio, and
>>         a Compact_Disk
>> 
>> If you want to describe something (an audio book) that is available in
>> several formats, you are describing relationships between different things.
>> 
>> Against my better judgement and dipping into FRBR language to explain
>> it....
>> 
>> You would have the description of an Expression, of type Audiobook, with
>> links to instances (Manifestations) for each format. Each instance would
>> be a combination of Audiobook and Compact_Disc; Audiobook and DVD;
>> Audiobook and Cassette; etc.
>> 
>> Check out the examples library
>> A0<http://www.w3.org/community/schemabibex/wiki/Examples/mylib/A0>
>> (Expression) and its related instances (Manifestations)
>> A1<http://www.w3.org/community/schemabibex/wiki/Examples/mylib/A1> and
>> A3<http://www.w3.org/community/schemabibex/wiki/Examples/mylib/A1> to
>> see how this might be encoded.
>> 
>> 
>> 
>> Moving on to how we encode multiple types for a thing there are a couple
>> of issues to address.
>> 
>> Firstly, the differences between RDF (Turtle), RDFa, and Microdata.
>> 
>>  * RDF is the most obvious – as per the above example you just keep
>>    adding type statements as required.
>>  * RDFa add the type URI to the ‘typeof’ attribute:
>> 
>>        <div vocab="http://schema.org/"
>>             typeof="Audiobook
>>    http://www.productontology.org/id/Compact_Disk">
>> 
>>  * Microdata is a little more difficult as the microdata standard does
>>    not natively support multiple types.  To overcome this limitation
>>    Schema introduced the addtionalType property so that they could
>>    encode this concept using microdata, thus:
>> 
>>        <div itemscope itemtype="http://schema.org/Audiobook">
>>             <link itemprop="additionalType"
>>    href="http://www.productontology.org/id/Compact_Disk">
>> 
>> 
>> The fact that the microdata solution uses additionalType as the property
>> name introduces the impression that the other type(s) are somehow
>> subordinate.  Maybe it would have been better to have ‘alsoOfType’ as a
>> property name.
>> 
>> The important effect of this approach is that there is no relevance in
>> the order of their declaration.  For instance a librarian may describe
>> an audiobook on CD in microdata thus:
>> <div itemscope itemtype="http://schema.org/Audiobook">
>> 
>>      <link itemprop="additionalType"
>>    href="http://www.productontology.org/id/Compact_Disk">
>> 
>> 
>> Whereas a retailer may describe the same thing as:
>> <div itemscope itemtype="http://www.productontology.org/id/Compact_Disk">
>> 
>>      <link itemprop="additionalType" href=" http://schema.org/Audiobook">
>> 
>> These are both valid and equivalent to each other.
>> 
>> 
>> ~Richard
>> 
>> 
>> On 09/02/2013 20:09, "Karen Coyle" <kcoyle@kcoyle.net> wrote:
>> 
>>    Owen, I take your point: additionalType seems to be sub-typing
>>    CreativeWork, not adding information about the product. I vaguely recall
>>    having been warned about additionalType -- that it is not often used and
>>    seems to be tricky. Here's the definition of "aT":
>> 
>>    "An additional type for the item, typically used for adding more
>>    specific types from external vocabularies in microdata syntax. This is a
>>    relationship between something and a class that the thing is in. In RDFa
>>    syntax, it is better to use the native RDFa syntax - the 'typeof'
>>    attribute - for multiple types. Schema.org tools may have only weaker
>>    understanding of extra types, in particular those defined externally."
>> 
>>    Richard posted this in an email: [1]
>>    "
>>    Sticking with the Product Ontology approach for a moment ­ an audiobook
>>    in WMA on a cd would just be a combination of multiple types thus:
>>      > http://schema.org/Book
>>      > additionalType: http://www.productontology.org/id/Audiobook
>>      > additionalType: http://www.productontology.org/id/
>>    Windows_Media_Audio
>>      > additionalType: http://www.productontology.org/id/ Compact_Disc
>>      >
>> 
>>    First, I think that "/associatedMedia" in CreativeWork looks to be a
>>    better fit for this. It is defined as: "The media objects that encode
>>    this creative work. This property is a synonym for encodings."
>> 
>>    Second, it still isn't clear, however, if you have multiple
>>    associatedMedia fields, e.g. A, B and C, whether that means that you
>>    have that CW in three different media, or if you have the CW in a single
>>    medium that is defined as A+B+C. I believe that Richard's example above
>>    was the latter. You seem to be concerned about encoding the former.
>>    Surely we need to be able to distinguish between them. I believe that
>>    means moving toward item or offer-level description for the different
>>    encodings. I can't think of any other way to make it clear.
>> 
>>    kc
>> 
> 
> -- 
> Karen Coyle
> kcoyle@kcoyle.net http://kcoyle.net
> ph: 1-510-540-7596
> m: 1-510-435-8234
> skype: kcoylenet
>