[foaf-dev] Proposed changes to Relationship vocabulary

[Damian Steer]

Danny Ayers wrote:
> 2008/7/4 KANZAKI Masahide <mkanzaki at gmail.com <mailto:mkanzaki at gmail.com>>:
> 
>     Hi Ian,
> 
>     A subproperty of a symmetric property not necessarily be symmetric. It
>     can be, but isn't entailed to be.
> 
> 
> Of course I trust your statement on this, but would be grateful for a 
> pointer the appropriate part of the specs where it appears (I'm overdue 
> a total re-read of the whole spec suite, but there are rather a lot and 
> I'd completely forgotten this part).

As has been said, the specs don't enumerate things which aren't true, 
but you can work through the semantics to see why this fails:

x subpropertyof y: extension of x is a subset of the extension of y.
z symmetric property: if <a,b> in extension of z then <b,a> is, too.

Let's take a property P, extension { <a,b> <b,a> }.

Obviously symmetric.

Take a property Q, extension { <a,b> }.

Obviously a subproperty (EXT(Q) subset of EXT(P)), and obviously not 
symmetric.

Convincing enough?

Damian