W3C home > Mailing lists > Public > public-vocabs@w3.org > August 2013

Re: Proposal: Looking inside tables

From: Omar Benjelloun <benjello@google.com>
Date: Thu, 15 Aug 2013 11:12:07 -0400
Message-ID: <CACsq2mkfpLmhkBmQ6mZE-17p7pGAhMuK1EN9FKaWAq3JjCoWzw@mail.gmail.com>
To: Thad Guidry <thadguidry@gmail.com>
Cc: Markus Lanthaler <markus.lanthaler@gmx.net>, Stéphane Corlosquet <scorlosquet@gmail.com>, W3C Vocabularies <public-vocabs@w3.org>, Ramanathan Guha <guha@google.com>, Dan Brickley <danbri@google.com>
Hi Thad,

The type declarations should all use SetOf, whether they're in
schema.orgor class URLs from another vocabulary.

The @id definitions (and other properties) are patterns where the values of
columns will be substituted for each row, to generate the values of the
properties for each instance, so the SetOf substitution doesn't apply to
them.

Does that help clarify?

Thanks,
-Omar


On Thu, Aug 15, 2013 at 10:39 AM, Thad Guidry <thadguidry@gmail.com> wrote:

> Omar,
>
> So the /SetOf/ will be optional in user defined urls or not ?  The
> examples under "Identifers and references" section seem to strip it out, or
> I misunderstood your earlier email :
>
>   "@type": "SetOf/Country",
>
>   "@id": "http://my.domain.org/country/{t1:col:country-code}",
>
>
> On Thu, Aug 15, 2013 at 9:21 AM, Omar Benjelloun <benjello@google.com>wrote:
>
>> Attached is a new version of the proposal that uses SetOf as a way to
>> declare table annotations explicitly (in the HTML mark-up as well as in
>> JSON-LD)
>>
>> @Markus we also considered explicit manipulation of tables in an earlier
>> version of this proposal. I'll try to explain why we went with the current
>> approach:
>>
>> If you look at the examples towards the bottom of the document, you'll
>> see that for the more complex cases, the directionality of the mappings is
>> different: instead of mapping from table columns to properties of a type,
>> we map from the type and its properties (sometimes with nesting) to columns
>> / patterns. This is more expressive, as the graph schema structured can be
>> nested, while the table structure is flat.
>>
>> Mapping from the type structure to the table columns with an explicit
>> table layer becomes quite heavy, as you first need to define the tables and
>> their columns, then the type/property structure with references to the
>> columns.
>>
>> I think the current approach with explicit SetOf is a nice trade-off
>> between simplicity and expressiveness.
>>
>> Thanks,
>> -Omar
>>
>>
>>
>> On Thu, Aug 15, 2013 at 9:55 AM, Markus Lanthaler <
>> markus.lanthaler@gmx.net> wrote:
>>
>>> On Wednesday, August 14, 2013 8:32 PM, Stéphane Corlosquet wrote:
>>> > From the proposal:
>>> >
>>> > <table typeof="Painting" vocab="http://schema.org/">
>>> >   <thead>
>>> >     <tr>
>>> >       <th property="image">Image</th>
>>> >       <th property="name">Title</th>
>>> >       <th property="dateCreated">Year</th>
>>> >       <th>Technique</th>
>>> >       <th>Dimensions</th>
>>> >       <th property="contentLocation">Gallery</th>
>>> >     </tr>
>>> >   </thead>
>>> > <tbody>...</tbody>
>>> > </table>
>>> >
>>> > It should be noted that parsers which are not aware of this table
>>> > extension would generate this information:
>>> >
>>> > <>
>>> >    rdf:type schema:Painting;
>>> >    schema:image "Image";
>>> >    schema:name "Title";
>>> >    schema:dateCreated "Year";
>>> >    schema:contentLocation "Gallery" .
>>> >
>>> > How do you plane to cope with this situation? Leave it be? Would it
>>> > have unintended consequences on some applications?
>>>
>>> Thanks Stéphane, I was going to ask the same question. I think the
>>> problem is even more apparent if you look at the JSON-LD examples in the
>>> draft:
>>>
>>> {
>>>   "@context": "http://schema.org/",
>>>   "@type": "Painting",
>>>   "dateCreated" : "{http://wp.org/rembrandt-paintings.csv#col:Year}",
>>>   "contentLocation" : "{
>>> http://wp.org/rembrandt-paintings.csv#col:Gallery}",
>>>   "author": "http://en.wikipedia.org/wiki/Rembrandt"
>>> }
>>>
>>> Why not define something which is a bit more explicit at the price of
>>> being a bit more verbose? Something like
>>>
>>> {
>>>   "@context": "http://schema.org/"
>>>     "rp" : "http://wp.org/rembrandt-paintings.csv#"
>>>   },
>>>   "@type": "Table",   ----- in lack of a better name
>>>   "columnDefinitions": [
>>>     {
>>>       "@id": "rp:col:Title",
>>>       "mapsTo": "http://schema.org/name" -- can also be abbrev. to
>>> "name"
>>>     },
>>>     {
>>>       "@id": "rp:col:Year",
>>>       "mapsTo": "dateCreated"
>>>     {
>>>       "@id": "rp:col:Gallery",
>>>       "mapsTo": "contentLocation"
>>>     {
>>>       "constants": {
>>>         "@type": "Painting",
>>>         "author": "http://en.wikipedia.org/wiki/Rembrandt"
>>>       }
>>>     }
>>>   ]
>>> }
>>>
>>> Or in HTML
>>>
>>> <table typeof="Table" vocab="http://schema.org/">
>>>   <thead rel="columnDefinitions">
>>>     <tr>
>>>       <th about="#image" property="mapsTo"
>>> resource="schema:image">Image</th>
>>>       <th about="#title" property="mapsTo"
>>> resource="schema:name">Title</th>
>>>       <th about="#year" property="mapsTo"
>>> resource="schema:dateCreated">Year</th>
>>> ...
>>>
>>>
>>> Cheers,
>>> Markus
>>>
>>>
>>> --
>>> Markus Lanthaler
>>> @markuslanthaler
>>>
>>>
>>>
>>>
>>>
>>>
>>
>>
>> --
>> Omar Benjelloun | benjello@google.com | (415) 845-8516
>>
>
>
>
> --
> -Thad
> Thad on Freebase.com <http://www.freebase.com/view/en/thad_guidry>
> Thad on LinkedIn <http://www.linkedin.com/in/thadguidry/>
>



-- 
Omar Benjelloun | benjello@google.com | (415) 845-8516
Received on Thursday, 15 August 2013 15:12:58 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:29:29 UTC