W3C home > Mailing lists > Public > public-vocabs@w3.org > April 2014

Re: Socialnetworks of a person or organization

From: Dan Brickley <danbri@google.com>
Date: Wed, 9 Apr 2014 17:09:42 +0100
Message-ID: <CAK-qy=5QefSMR2pKFAE+X8Ya-WbzbEkmFFW0yem4J8ZxuMrhGg@mail.gmail.com>
To: "martin.hepp@ebusiness-unibw.org" <martin.hepp@ebusiness-unibw.org>
Cc: Jason Douglas <jasondouglas@google.com>, David Deering <david@touchpointdigital.net>, W3C Web Schemas Task Force <public-vocabs@w3.org>
On 9 April 2014 16:53, martin.hepp@ebusiness-unibw.org
<martin.hepp@ebusiness-unibw.org> wrote:
> I think that Dan's rationale is, and I support that, that we can assume that someone using the Wikipedia / Freebase URI of a page about an entity typically means the entity, not the page.
>
> So we take the URI of the Wikipedia page about John Lennon as an identifier for John Lennon.
>
> I think we can leave the rathole of solving the disambiguation problem between an entity and a representation of the entity out of the level of data representation, and the whole HttpRange-14 [1] problem behind, and assume that the clients consuming data will be able to make that distinction from contextual information and other signals.

The schema.org/sameAs property definition tries to sneak around those
ratholes by explicitly saying that a Web page / url is expected
(contrast with owl:sameAs which is true numerical identity).

Back to (social)account, the other issue I mentioned is that for
complex organizations with many associated accounts (even 'official'
ones, never mind fans/friends/fake...) there could be several
candidate associated or 'subsidiary' entities.

Consider for e.g. the BBC as an entity, and this little collection of
Twitter accounts I just searched for: @BBCBreaking, @bbc5live,
@BBCTech, @BBCClick, @BBCSport, @bbcworldservice, @BBCtrending,
@bbcdoctorwho, @BBCRecruitment, @AboutTheBBC. Some of these are
presented as team accounts, some are for a TV show or station or topic
or activity or brand. Schema.org has the ability to make lots of
matching fine-grained distinctions (see
http://blog.schema.org/2013/12/schemaorg-for-tv-and-radio-markup.html),
but we wouldn't want to end up implying - for example - that Doctor
Who (a http://schema.org/TVSeries) was the same entity as BBC
Recruitment (considered as a 'department' of the 'Organization' that
is the BBC).

That seems to me to be the strongest argument for an 'account'
property, rather than recycling 'sameAs'. But it was worth trying to
squeeze more usage out of sameAs and there are some subtle
inter-relationships here. It ought to be reasonable to have a property
linking the BBC to each of these (assuming they're appropriately
official), even if they could also be linked to teams, shows, brands,
departments, web sites, etc etc.

That said, a single schema.org property is unlikely to capture all the
nuance here. It might be that Person descriptions will work better
with a single property description than complex organizations.

Dan




> Martin
>
> [1] https://www.w3.org/2001/tag/group/track/issues/14
>
>
> On 09 Apr 2014, at 17:27, Jason Douglas <jasondouglas@google.com> wrote:
>
>> Adding to that... if the claim is that authorship or control should be verified out of band, then one of the common mechanisms for doing that is bi-directional links.  If we lump everything under sameAs, we could end up with bidirectional links for both the Wikipedia page and the Facebook profile.
>>
>> On Wed Apr 09 2014 at 8:24:31 AM, Jason Douglas <jasondouglas@google.com> wrote:
>> -1  There's a difference between reference pages *about* the same entity and pages authored/controlled *by* the same entity.
>>
>> On Wed Apr 09 2014 at 8:18:36 AM, David Deering <david@touchpointdigital.net> wrote:
>> Sounds good to me, Dan.  It would make things simpler and cleaner.
>>
>>
>> David
>>
>>
>>
>> On 4/9/2014 10:11 AM, Dan Brickley wrote:
>>> Revisiting this and the recent socialAccount thread,
>>>
>>> http://lists.w3.org/Archives/Public/public-vocabs/2014Apr/0046.html
>>>
>>>
>>> There seems to be broad agreement that it would be good for
>>> schema.org
>>>
>>> to recommend a pattern for marking up links to (broadly) social
>>> network profile pages, e.g. Twitter. However a few people have raised
>>> the concern that adding another property will add more confusion
>>> around existing options, particularly 'url' and 'sameAs'.
>>>
>>> Therefore, a minimalistic revised proposal: that we address this
>>> scenario using 'sameAs' directly.
>>>
>>> I suggest
>>> http://schema.org/sameAs
>>>
>>> "URL of a reference Web page that unambiguously indicates the item's
>>> identity. E.g. the URL of the item's Wikipedia page, Freebase page, or
>>> official website." therefore becomes
>>> "URL of a reference Web page that unambiguously indicates the item's
>>> identity. E.g. the URL of the item's Wikipedia page, Freebase page, a
>>> profile page on a social site, or official website."
>>>
>>> We should also add examples at least for Person, Organization and
>>> MusicGroup to illustrate this.
>>>
>>> Following the example in the wiki at
>>>
>>> https://www.w3.org/wiki/WebSchemas/SocialAccountProperty
>>>  the usage
>>> then would be something like:
>>>
>>> <div itemscope itemtype=
>>> "http://schema.org/Person"
>>> >
>>> <span itemprop="name">Stephen Fry</span>
>>>     (<a itemprop="url" href=
>>> "http://www.stephenfry.com/">stephenfry.com
>>> </a>,
>>>      <a itemprop="sameAs" href=
>>> "http://twitter.com/stephenfry"
>>> >twitter</a>,
>>> <a itemprop="sameAs"
>>> href=
>>> "http://en.wikipedia.org/wiki/Stephen_Fry">wikipedia</
>>> a>)
>>> </div>
>>>
>>> This has the advantage of not requiring the (endlessly evolving and
>>> slippery) notion of 'social' to be defined. Or 'account' for that
>>> matter. It removes some worry for publishers, "am I using
>>> 'socialAccount' when I should be using 'sameAs' or vice-versa?). It
>>> carries a little less meaning, but not a lot. Someone writing an app
>>> to find twitter links will know just what they need to do. If your
>>> goal is to sort 'social' from 'other kinds of authority page', you'll
>>> need out-of-band information of some kind. But that was likely also
>>> going to be the case even if we added a new property 'socialAccount'.
>>>
>>> How does this sound?
>>>
>>> Dan
>>>   (sameAs
>>> <http://twitter.com/danbri/>
>>> )
>>>
>>> p.s. just a reminder,
>>> schema.org
>>> 's notion of sameAs allows for
>>> identity reference pages as values, e.g. hints for entity
>>> identification. It does not mean 'numerical identity', i.e.
>>> self-same-thing; for that you could use owl:sameAs.
>>>
>>>
>>
>>
>
>
Received on Wednesday, 9 April 2014 16:10:50 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:29:39 UTC