Re: Actions in Schema.org: target & object properties

Good points. Answers within.

On Thu, May 16, 2013 at 5:00 PM, Jim Klo <jim.klo@sri.com> wrote:

> Thanks Yaar,
>
> Certainly makes more sense. My only concern is that this approach may
> generally require domain specific extension of Schema.org.  I'm not sure
> that's a good or bad thing, however at onset I'm not super comfortable with
> the approach, only that discovery of DSL extensions can prove to be
> cumbersome.  My use cases in LR are really not much different than the
> general internet indexing in that anyone can publish anything… so having
> more well specified core vocabularies help.
>

The Schema.org Actions proposal tries to cater to the millions of
publishers & developers who would hopefully define actions and action
handlers on their websites, emails and apps. They need precise standards to
collaborate with each other, and they need the vocabulary to be expressive
enough to capture domain-specific nuances. For example, AcmeCarRentals.com,
who rents cars in the San Francisco area, doesn't just want to know that
you (actor) want to rent (verb) car (object). It also needs to know what
type of car you are interested in, miles you plan to drive, pick up and
drop off location, etc.

By standardizing the verb modifiers, not just the verb names, Schema.org
Actions is designed to reduce complexity for the many publishers who only
need to deal with 1-10 action types, at the expense of potentially
increasing complexity for the aggregators, who might need to deal with
automatic discovery (or maybe not - see my next point).


> FWIW, as an aside - I have my doubts on the whole Schema.org extension
> process which just feels haphazard.  To me I just see the flurry of
> http://schema.org/Action/FubarAction appearing without really
> understanding how to process the properties.
>
I'm concerned about that too. But the flurry would be controlled by how
many people process and do something meaningful with FubarAction. If nobody
understands it's properties, it won't be widely used. So the incentives to
stick to the standard actions is strong. This has proved to be the case
with other Schema.org non-action types.

I hope schema.org will move fast to introduce an extensible, useful
vocabulary of verbs. That will reduce the need to create FubarActions.


> The other bit with your examples is that I'm tending to disagree with is
> of the "need" for specialized object.  To me it seems like this
> specialization make the object more Human readable than Machine readable.
>  Unfortunately I don't know the full history of how you got this far in the
> proposal, hence I'm blind to the driving decisions that were made.  But I'd
> argue that you could have a more generalized action and still accommodate
> some modifiers.  It wouldn't negate the ability to specialize, I just see a
> world where extension of Action is a norm if there is not a generic -
> albeit not perfect - solution available.
>
Schema.org takes the approach that everything is a Thing, but for people to
collaborate, you need to standardize what sub-Things are possible and which
properties each of them has, using a type system. In Schema.org Actions, we
simply extend this successful methodology to verbs as well. There are other
ways to standardize verbs and their properties, but the current approach
worked for other types, and there is value in consistency.


>
> On May 16, 2013, at 4:12 PM, Yaar Schnitman <yaar@google.com>
>  wrote:
>
> Jim, you asked about adding "target" and "object" properties to
> schema.org/Action.
>
> Taking note of ActivityStrea.ms, we certainly considered adding "target"
> and "object" to the Schema.org Actions proposal. However, we found that
> such properties will create problems if our goal is to be able to fully
> express actions in very precise and unambiguous ways.
>
> Furthermore, relying on "target" and "object" also becomes a challenge in
> the case of "potential" actions, where the "target" and "object" might
> still be missing, since not decided yet.
>
>
>
> Not sure this is true… "potential" actions in the context of say a web
> form:
>
> AuthenticatedUser (actor) can post (tentative action)  FormX  (object) to
> SystemY (target)
>

>
> In a "potential action" use case, the "performer" could be described as
> the type of permissions needed to execute the action.
>
> All actions must have an object that is being acted upon… it may be
> inferred in the though the language:
>
> "Yaar drove."  -- If you drove… you drove some "Thing"… its just not well
> described, but could be inferred by context.
> "Yaar can drive."  -- Indicates potential driving… the what you can drive
> is still a "Thing"…
>
> And really to be honest… I think most people would tend to use
> Thing.action:
>
> SnazzyNewWidget.action [ Buy, Bid, Comment, Bookmark, Favorite, Share ]
>

> As I think the larger use case of this vocabulary is I'm searching for
> "Snazzy New Widget" and the search index returns results with actions to
> Buy, Bid, Comment, Bookmark, Share, etc...
>

To give you some real-world context, the proposal has evolved in the
context of publishing actions over email for Gmail. It was launched
yesterday.


>
> - JK
>
>
>
>
>