[rdfweb-dev] Relationship Schema Updated

Michael Bauser michael at bauser.com
Fri Mar 19 04:30:25 UTC 2004

Hash: SHA1

Richard Lennox wrote:

> Ok Im pretty new to this but will throw my opinion into the pot.
> Firstly, is it not enough to just say spouseOf? There is no indication
> of wife or husband just that  X is the spouse Of Y.  In todays
> societies there is no reason as to why gender should come into this
> relationship at all. Gender describes the person not the
> realationship.

Actually, "sex" describes the person. "Gender" describes a role. The two
became confused because Supreme Court Justices didn't like using the
word "sex" in their opinions.

> The point regarding male-roles and female-roles in the relationship -
> maybe 50yrs ago there were defined roles but not today.

Um, I think that was my point, too. I think you're seeing a moral
judgement where there's just meant to be data. (And perhaps making a
moral judgement of your own, about what it means to recognize a difference.)

I'm not advocating things like "mother" and "father" as designations of
anatomy, or even social duties. (RDF doesn't care who does the
housework, I hope.) "mother" and "father" here are coordinates in a
social network that are used *with* other descriptive data (cultural
background, bloodline, and temporal data, mostly) to infer social
information. (If for some reason you needed to write a vocabulary that
says "only mothers cook dinner", go for it. Just keep that out of the
base data.) Take, for instance, the relatively trivial example of
"inferring what words I use to describe my relatives". Here are some my
first- and second-degree kin, along with the English terms I use to
describe them:

Debra		mother
Gail		aunt
Heather		cousin
James		uncle
Jeff		cousin
Kristin		cousin
Linda		aunt
Matthew		cousin
Michael		father
Raquel		cousin
Robert		cousin

You get, I get it, everyone here gets it. OK. Now, if for some reason,
you thought I was one of the Grand Valley Dani of New Guinea:

Debra		akoja
Gail		he opaije
Heather		akosa
James		ami
Jeff		ejak
Kristin		ejak
Linda		akoja
Matthew		oe
Michael		opaije
Raquel		oe etu
Robert		ami

See that lack of one-to-one correlation? That's because English kinship
terms are actually atypical in how little information they use. It's
unusual (in the overall scheme of things) that the English system
doesn't care what "side of the family" my kin is on: most systems
distinguish between the two sides for at least some kin terms. (If you
had more data about how I'm related to my aunts and uncles, for example,
you might see how the Dani system makes different distinctions for
male-male-male connections (generation is more important than sex)
than it does about female-male-male connections (sex is more
important than generation), and that might even lead you to some
insights about how the Dani trace lineages. Kinship is about the *path*
from relative to relative, which in the case of the Dani, is
anatomy-influenced, because Dani kinship doesn't acknowledge
gender-changes. In the case of Nuer, and apparently the family of
everybody on this list, it's not always anatomy-influenced because they
*do* recognize gender change.)

And that's the *trivial* example. In more elaborate scenarios, those
paths provide guidlines for things like who to invite to my wedding or
funeral. That's interesting data, and it might even be useful to any
Dani members who are planning my funeral.

And again, while many societies define "side" by anatomy, not all of
them do. Even the societies that remove anatomy as a qualification for
"motherhood" or "fatherhood" recognize the difference between "mother"
and "father" *because* they need it as part of the coordinates in a
kinship network. The problem with the proposed relationship vocabulary
is that it *intentionally* throws away data relevent to *more than half*
the people in the world *without* admitting its biased. "Generic
parents" break most kinship systems. It's a very biased vocabulary that
will make it harder for people in different societies to understand each
others' data.

> Other relationships can be inferred through logic.  parentOf for
> example, gender specifies that the person is either father or mother,
> despite the practical parental role played in the childs life.

You missed that foaf:gender is not limited to male/female. If you want
FOAF to describe anatomical sex, then somebody has to propose
"foaf:genitals". I'm not ready to do that.

> For example, scientifically at this point in time

"this point in time"? Ouch. We're back to cultural biases. I'm trying to
advocate a vocabulary that's flexible enough to store data about social
networks past, present, *and* future. If you're thinking we only have to
describe what "we" do "now", you're letting a subtle cultural bias lead
to a bad design decision: You're advocating a fragile standard that you
*know* will eventually break.

Also, anthropology *is* a science.

> Women cannot be fathers (maybe with gene work in the future but there
> would need to be great social change).

We're mixing up two things here: *kinship* is the description of social
relationships; *pedigree* is the description of biological
relationships. My concern has been primarly about the social network
(which is why I didn't originally propose any distinctions between
"biological father" and "adoptive father", but I see now that won't
quite work for the masses). When it comes to describing social networks,
it come down to this: If I ask somebody "Who is your father?" and they
point to someone I'm *sure* is a woman, it doesn't mean that woman isn't
a father; it means that the person I asked a question of doesn't require
fathers to have penises.

If you want to describe a *biological* network of pedigree-relations,
you either need a separate vocabulary, or an additional property (I'm
now leaning towards biological/marriage/other as a second relationship
property to be used *with* the eight relationships, as I explain in my
reply to Jim.)

(A "pedigree vocabulary" would probably be even smaller than a kinship
vocabulary, since spouse/husband/wife isn't really relevant. It will
eventually need a relationship for "clone", though.)

> Somewhat different point: should there not be something to say while
> not biologically parentOf, parentOf in the case of adoption.

Probably, but the other vocabulary refuses to deal with that, too. I
*believe* it can be handled (without too much difficulty) by using the
second property, and/or a reference to a life event (for instance if the
FOAF file includes a description of an "adoption event", any foaf:Person
linked to that event would presumably be the adoptor or the adoptee.)
Using a few good events is requires less new vocabulary than creating
kin:adoptiveFather, kin:adoptiveMother, kin:adoptiveBrother, and so on.
Version: GnuPG v1.2.1 (MingW32)


More information about the foaf-dev mailing list