[sc34wg3] New syntax for (binary) associations

[Dmitry]

On 3-Feb-08, at 7:17 AM, Steve Pepper wrote:

> * Robert Barta
> |
> | **IF** we go for a default role type, then I would prefer
> | tm:thing (subject, whatever) for both. 'Subject' and
> | 'Object' describe the roles of something _in the statement_,
> | not what I actually want to express.
>
> This is the point I was trying to make about how Dmitry is
> conflating grammatical roles and semantic roles. Subject and
> object belong to the realm of syntax. In semantics one finds
> other role types, such as agent, experiencer, patient, theme,
> etc. In any given utterance each entity plays one semantic role
> and one grammatical role.
>

We are working at the level of representations.  "is-employed-by 
(X,Y)" can represent exactly the same thing as "is-employed-by 
(employee: X, employer: Y)"
I think it is enough to have a good text description which explains  
meaning of "is-employed-by(X,Y)" . After that we can use compact form.



> What we need is the ability to say that in a particular context,
> role type X (say, employee) is to be regarded as the "subject",
> and role type Y (say, employer) is to be regarded as the
> "object". The same mechanism could in theory be used to annotate
> an ontology in order to support automated translation to RDF.
>
> So instead of tm:subject and tm:object we need something like
> tm:subject-role and tm:object-role.
>

I like the  idea of explicit "Property" definitions

o:works-for
     isa Property;
     tm:subject_role o:Employee @  o:is-employed;
     tm:object_role o:Employer @  o:is-employed.

Explicit properties allow to have a compact syntax for binary  
associations + some new interesting possibilities:


john
     meta:unknown      o:works_for;
     meta:known_all    o:parents.


Properties  also provide  ability to express TMDM instances  in  
compact form (when Property definitions are available).

Dmitry