Re: schema.org action status type

It appears that JSON-LD is being used to serialize activity stream 
information, e.g., in 
http://jasnell.github.io/w3c-socialwg-activitystreams/activitystreams2.html
But then the vocabulary is stated in terms of classes and terms in
http://jasnell.github.io/w3c-socialwg-activitystreams/activitystreams2-vocabulary.html
This produces a strong disconnect with the formal model underlying JSON-LD - 
RDF.  It makes is very hard to figure out just what is allowed, expected, and 
forbidden.

Some of the vocabulary is very jarring.  For exammple, what is a "Natural 
Language Value"?  Is is something that has as a value a string in some natural 
language?  Why not directly use strings with language tags, instead of writing 
a lot of stuff that is hard to understand?  Similarly with Links.  Why spend 
so much time talking about something that is build in to RDF?

You have actions and potential actions, which appear to share the same set of 
properties.   This opens up a large can of modelling terms, as potential 
actions describe a category of actions, and thus deserve to be treated as 
classes, not individuals.

This is only a quick list of things that jumped out at me.  There might be 
more modelling issues in the proposals that are only evident on closer 
examination.  Of course, modelling is a never-ending task, and you do need to 
do something.  You should however try not to inadvertently make your modelling 
problematic.


If your model fits within RDF, then why not use RDF throughout.  Turtle is a 
nice syntax for RDF, and would make the examples closer to the model.  If your 
model doesn't fit within RDF, then you might consider using OWL as the 
modelling language.  If your model doesn't fit within OWL, then ....

peter

On 11/03/2014 07:58 AM, James M Snell wrote:
> Much of that is due to purposeful lack of a strong semantic definition in the
> original Activity Streams 1.0 spec. For 2.0, this kind of feedback is helpful,
> however. If you'd be so inclined, some additional notes on what information
> and definition may still be required would be helpful (some of us working on
> it are so intimately familiar with the spec that it's often difficult to take
> that kind of neutral objective view)
>
> - James
>
> On Sun Nov 02 2014 at 11:26:51 AM Peter F. Patel-Schneider
> <pfpschneider@gmail.com <mailto:pfpschneider@gmail.com>> wrote:
>
>
>
>     On 11/02/2014 10:16 AM, James M Snell wrote:
>      >
>      > On Nov 2, 2014 9:48 AM, "Peter F. Patel-Schneider"
>     <pfpschneider@gmail.com <mailto:pfpschneider@gmail.com>
>      > <mailto:pfpschneider@gmail.com <mailto:pfpschneider@gmail.com>__>> wrote:
>      >  >
>     [...]
>      >
>      >  > One very strange aspect of these documents is that they appear to be
>     trying
>      > to create a formal model for activities.  However, they use quite a
>     variety of
>      > semi-formal concepts for this purpose.  Why not just use a formal modelling
>      > system for this purpose?  You wouldn't have to use RDF or OWL if these
>     systems
>      > are in a great disfavour in the working group.
>      >  >
>      >  >
>      >
>      > How is that strange? Especially given that defining a model for
>     activities is
>      > precisely what the activity streams document does ;-). So far, a formal
>      > modeling system has not been necessary but some steps have been taken
>     in that
>      > direction.
>
>     Well the strangeness is that it is very hard to figure out just what is
>     supposed to be going on because the foundations are underdefined.
>
>      > - James
>
>     peter
>

Received on Monday, 3 November 2014 16:22:17 UTC