The perils and delights of multiple type entities for schema.org (was Re: Proposal: schema.org/MobileVideoGame)

   1. AFAIK, information for webmasters on how to express multiple type
   entities (hereafter MTEs) is non-existent.  There's no documentation on the
   subject on schema.org on itself nor, to the best of my knowledge, on any
   of the sponsor's sites (neither Bing, Yahoo, Yandex nor Google provide
   guidance in the use of MTEs).

   The only way I've learned anything at all about correct MTE syntax is
   via this mailing list, which is hardly a primary resource for most
   webmasters.

   All of this makes MTE use difficult for most webmasters, and increases
   the chance that when MTEs are employed they'll be employed incorrectly.

   2. Search engines do not provide visible support for MTEs.

   I use the word "visible" because its by no means clear whether or not
   the search engines correctly support MTE declarations, nor *which* search
   engines understand MTEs.

   The new Google Structured Data Testing Tool doesn't seem to support MTEs
   well, and the example posted by Jarno van Driel a year ago still returns
   errors that shouldn't be present if the MTE declarations were understood
   correctly.  The Yandex Structured Data Validator throws an error for the
   exact code provided for the mobile video game example I previously
   referenced (from schema.org/VideoGame).

   And while these are errors reported by specific data consumers, this
   should concern schema.org IMO, and not just because the engines
   referenced are schema.org sponsors.  These tools - and especially the
   Google SDTT - are touchstones for webmasters wishing to employ structured
   data, and when good markup requires the use of an MTE *and then* throws
   a validation errors, webmasters are liable to produce inferior markup by
   eradicating a type, or by abandoning their structured data efforts.

   This does not support "the purpose of schema.org" to "improve the web by
   creating a structured data markup schema supported by major search
   engines", [2] especially when the major search engines give every
   indication that they don't support the structured data markup webmasters
   are being encouraged to use.

   3. Use of MTEs makes it more difficult to extend the use of schema.org
   (which the initiative also implicitly acknowledges as a goal in saying that
   "Markup can also enable new tools and applications that make use of the
   structure").

   Would the sort of applications that have been worked on (e.g. [3],[4])
   function with the additional complexity introduced by MTEs?  Don't know,
   but they *do* add complexity.

None of this is to say the use of MTEs isn't occasionally, or even often,
warranted, and certainly not to say that they shouldn't be used.

But it is to say that, as Martin Hepp has previously said on the subject
[5] (and he's also enumerated some of the syntactical challenges of MTEs
[6]), that there is "a trade-off decision" to be made here.  Yes, we don't
want to - and quoting Martin again here - define "dozens of subtypes or
additional siblings for existing types."  But it seems to me to be a
reasonable trade off to create a sub-type when the entity class is question
is widely represented on the web, and it is of no small significance in
these cases whether or not a data consumer can accurately understand the
data being provided - and I think mobile video games fall into that
category.

[1] https://lists.w3.org/Archives/Public/public-vocabs/2014Jan/0064.html
[2] https://schema.org/docs/faq.html#0
[3] http://moz.com/blog/semantic-analytics
[4]
http://apassant.net/2014/04/24/export-and-structure-your-musical-activity-with-schema-org/
[5] https://lists.w3.org/Archives/Public/public-vocabs/2014Oct/0173.html
[6] https://lists.w3.org/Archives/Public/public-vocabs/2014Jun/0052.html

On Fri, Feb 13, 2015 at 12:20 PM, Vicki Tardif Holland <vtardif@google.com>
wrote:

> Out of curiosity, what limitations do you see in having to use multiple
> types?
>
> - Vicki
>
> Vicki Tardif Holland | Ontologist | vtardif@google.com
>
>
> On Fri, Feb 13, 2015 at 3:07 PM, Aaron Bradley <aaranged@gmail.com> wrote:
>
>> schema.org v1.92 introduced the new type VideoGame, a more specific type
>> of both (the also-introduced) Game, and of SoftwareApplication.
>>
>> VideoGame is a great addition, but as mobile video games stand poised to
>> overtake console-based games in popularity [1], there is no way
>> differentiate between a traditional video game and this important variant.
>>
>> Freebase [2], Wikipedia [3] and Wikidata [4] all have entries for "mobile
>> game", and the Google distinguishes between "Video game" [5] and "Mobile
>> game" [6] in Knowledge Graph results generated on the basis of a video game
>> title search.
>>
>> Perhaps most tellingly, schema.org itself gives a nod to the importance
>> of mobile video games by providing a a markup example on the
>> schema.org/VideoGame page.  And in my opinion the way in which this
>> example is necessarily formulated demonstrates the utility of a
>> MobileVideoGame type:
>>
>> <script type="application/ld+json">
>> {
>>   "@context": "http://schema.org",
>>   "@type": ["VideoGame","MobileApplication"],
>>   "gamePlatform":"iOS",
>>   [...]
>> }</script>
>>
>> Only by means of this multi-type entity declaration is a data consumer
>> able to determine that a given video game is a mobile video game, and then
>> only by inference - for those data consumers that are able to correctly
>> process multi-type entities properly.
>>
>> In light of all of this, I propose MobileVideoGame, a more specific type
>> of both VideoGame and MobileApplication.  No additional properties would be
>> required to support this new type.
>>
>> Aaron Bradley
>> Electronic Arts
>>
>> [1] http://fortune.com/2015/01/15/mobile-console-game-revenues-2015/
>> [2] http://www.freebase.com/m/04951x
>> [3] http://en.wikipedia.org/wiki/Mobile_game
>> [4] http://www.wikidata.org/wiki/Q1121542
>> [5] https://www.google.com/search?q=battlefield%204&pws=0&hl=en&num=10
>> [6]
>> https://www.google.com/search?q=simpsons%20tapped%20out&pws=0&hl=en&num=10
>> [7]
>> https://developers.google.com/structured-data/testing-tool/?url=http://jsbin.com/niqile
>>
>
>