[rdfweb-dev] Another relationships proposal

Jim Ley jim at jibbering.com
Sat Mar 20 18:42:17 UTC 2004


"Danny Ayers" <danny666 at virgilio.it>
> Then I'd suggest there's the biological relationships - it could be
> (parent+gender) or (mother | father) . The latter seems a better bet,
> there is a cultural angle on gender. I would be tempted to include
> parent as a common superproperty - forget the cultural interpretation,
> just define rel:parent as *biological* mother or father.

Ignoring the cultural interpretation is silly, cultural interpretations are
what is important to the users, with FOAF we're generally talking about
self-asserted relationships of the community, we're not talking about
accurately modelling that community from outside for research or accuracy
purposes.  The users of rel:Parent as far as I can see care about being able
to say someone is their parent - revealing information about the exact
nature of that relationship (parent/AdoptedParent/Grandmother really etc.)
is not something that many individuals and cultures are happy with doing, it
reveal things which are not necessarily of value, and in some cultures carry
negative connotations - then there's the fact that some people may not know
the relationship.

Consider if an elder brother marks up their parent as not
rel:adoptiveParent, and the younger brother doesn't know about those
relationships - even the brothers refusal to make the assertion biological
may not be of value, like gender, interpersonal relationships matter to the
individuals more than the collective, and we should ensure that individuals
can say only what they want, and aren't constrained by our view of what it
should be.  It needs very, very general classes that can encompass
everything.

> I think those two relationships (motherOf, fatherOf) together with a bit
> of logic should be enough to allow modelling of all biological
> relationships.

Except of course I cannot model the relationship to my sister that model!
logic does not work in these relationships, labels such as sister, aunt etc.
matter.

> A brotherOf B
> =>
> X parentOf A
> X parentOf B

but that is not necessarily true, if we can't make such inferences then the
model is flawed.

Cheers,

Jim.




More information about the foaf-dev mailing list