[rdfweb-dev] Qualifying knows
wkearney99 at h...
Tue Dec 31 22:51:16 UTC 2002
> I agree with this. Knows is a fuzzy concept -- how well so I know Bill
> Kearney? We exchanged an email once. The degree of truth of "Doug
> knows Bill Kearney" is about 0.05.
> If I know Bill, and Bill Knows Jack, what can I infer about my
> relationship with Jack?
> - I can trust Jack less than I can trust Bill
> - I can trust Bill less than Jack trusts Bill
> I would suggest an Algebraic product for the fuzzy intersection AND.
> I trust Jack=(I trust Bill AND Bill Trusts Jack)
> anything else we can really infer from these?
Jen Golbeck's got some nice stuff going for trusts . I've put some trustweb
RDF in a test foaf . But that's not exactly what I'm talking about here.
It's not that I'm expecting this to be used for interpretations of trust.
That's an even more esoteric degree of the social aspects of grouping.
I don't see it being so much a matter of extendable trust. I see it more as
just a *single* level of association. Whether or not I know someone has nothing
to do with what someone else should or shouldn't infer. My knowing Ian and Doug
via e-mail by virtue of shared interest in things RDF has nothing to do with how
someone else might know them. But I wouldn't mind sharing this grouping. If
not as groups of individuals, as markup that indicated a foaf:knows individual
is known in a particular sort of way.
So I suppose I'm asking not only to markup the relationship, but to have a way
that foaf visualization tools might default to showing the relationships based
on qualifiers. Sort of a way to 'weight' the relationships such that if I
expressed "knowing of" someone it'd be shown with less 'value' than someone I
regularly interact with. The level or purpose of the interaction is a whole
other issue entirely. But basically, I want to be able to share the social
network information without making overly broad statements.
More information about the foaf-dev