[rdfweb-dev] Relationship Schema Updated

Michael Bauser michael at bauser.com
Fri Mar 19 04:30:49 UTC 2004

Hash: SHA1

Ian Davis wrote:

> Please explain for us novices why parent is less cross-cultural than
> father or mother? As far as I can see, it is a more general term for
> either.

By that definition of cross-cultural, "kin" is *really* cross-cultural,
but you're not advocating a generic "rel:kin" relationship, because
you've made a *value* judgement about which "general" relationships are
worth noting. Unfortunately, the generals you're advocating *aren't*
especially helpful, or even *basic*, because they don't combine well
enough to actually describe any kinship system *execept* English
bilateral kindreds. They're not good building blocks. It's a textbook
assumption of a cultural bias: You keep thinking your culture's system
is the one that everyone else can be fitted into.

Find me some cultures where people don't actually notice that they've
got two kinds of parents, and you'll have *half* an argument. (An
argument for leaving generic "parent" as a relationship; but I was going
to concede the need for that one anyway.) But you won't have anything
that proves using two parent-classes is biased, because it's not. It's
just descriptive.

> How do you use father/mother to describe foster relationships?

How do you use "Parent" to describe foster relationships? With a
qualifer of some sort, which is one of the things I've been working out
here, with helpful suggestions from other list members.

>>2) Kill FriendOf, CloseFriendOf, EnemyOf, AntagonistOf, and
>>AmibivilantOf. They're just going to create unhelpful ambiguities and
>>social awkwardness. (It's like an out-of-balance Likert scale, or
>>something.) If you really want "tiny steps", try
>>"postive/negative/neutral opinion" (with no assertions about
>>reciprocity). That's where a social scientist would start, because
>>informants alway understand binary questions.
> I think the existing terms are useful at the level FOAF is pitched. To
> help me understand your suggestion, are these the kinds of properties
> that you mean (forget the specific naming convention if you will)?
> positiveOpinionOf
> neutralOpinionOf
> negativeOpinionOf
> How would you qualify the opinion type? I may have a negative opinion
> of someone's ability to keep time, but a positive opinion of
> their loyalty.

I call "straw man" on that one. It doesn't make any sense.

If you want a vocabulary that deals with your opinion of somebody's
abilty to tell time or their loyalty, you *know* you can write the
"opinion of traits" vocabulary to avoid conflict with the relationship
vocabulary. Having rel:positiveOpinionOf to describe an overall opinion
of somebody doesn't prevent creating an
opinion:Don'tThinkHeCanTellTimeProperly, or even a rel:OpinionOfHisLoyalty.

FriendOf, CloseFriendOf, EnemyOf, AntagonistOf, and AmibivilantOf are
all just FeelingsTowardThisPerson in disguise. You've got a
positive/neutral/negative in the vocabulary already, it's just obscured
by lopsided design.

>>4) Lose the entire "Using With HTML and XHTML" nonsense. You're taking
>>link relationships in exactly the opposite direction the newer HTML
>>recommendations do; link relationships define the relationship between
>>documents, not people. You're imitating one of the worst aspects of XFN.
> This was explictly put in to embrace the XFN concept and to
> demonstrate that both camps can interoperate. I believe that if you
> specify the HTML profile you are using then the rel/rev attribute
> contents can be defined however you like.

I think that's a extremely iffy argument. The HTML 4.01 recommendation
is reasonably precise about the purpose of the "rel" attribute, and the
"link" element it's most often used with:

"rel This attribute describes the relationship from the current document
to the anchor specified"

"The LINK element defines a relationship between the current document
and another resource."

Section 12.3 of the Recommendation is, in fact, titled "12.3 Document
relationships: the LINK element".

Using rel to define a relationship between the current document and
non-document (like rel=Author, which is typically linked to an e-mail
address) is stretching the definition of "resource" a little, but it's
probably managable. Using it to define the relationships between two
non-documents is a *huge* stretch, and disrupts the map of the Web that
rel, rev, and link are supposed to create. You (and the XFN gang) are
advocating the disruption of somebody else's metadata system just to
promulgate yours. That's just plain tacky.

Metadata profiles are for defining relationships used by LINK and META;
using them to redefine the purpose of the HTML elements themselves isn't
their purpose. RDF vocabularies are not the right place to make changes
to how HTML works.

Besides, XFN is a disaster; it should be taken out and shot, not
encouraged or kissed up to.

> After all, it's only _a_ vocabulary for describing relationships. I
> think you're mistaking it for _the_ vocabulary which doesn't and
> cannot exist.

No, I'm mistaking it for the *first* vocabulary, which is going to set a
bad example for the ones that follow, by refusing to make its biases
explicit (and mucking up HTML). If it's really that important to you
that it only describe English-style families, the specification should
say "this vocabulary is only suitable for bilateral kindreds". That
would be start towards sanity.

Version: GnuPG v1.2.1 (MingW32)


More information about the foaf-dev mailing list