xml-dev - Semantic Web Hackings

["Sean B. Palmer" <sean@mysterylights.com>]

[Part of the "How to create a Semantic Web in just 10 days" series :-) ]
It occurred to me that I want a "real" Semantic Web presence, i.e.
something that says a little bit about me in RDF etc., but that is "future
proof": something that people using the SW in ten years (ahem) can come
across and gather useful information from.

Put more spuriously (and paraphrasing TimBL): "What's the WWW? I don't
know, but quick, we'd better have a WWW page or we'll get left behind!"
Now substitute SW for WWW, and you have the spirit of this note.

Onwards: I was wondering how you actually make assertions about a person
when there isn't a definitive namespace for a person on the Web. At first I
thought "we'll have to create one and then digitally sign it", but then I
remembered something the TimBL said at XML2000:-

     "if two people have the same email address you can be pretty
     certain that they are in fact the same person. Otherwise they
     have to sit on each others lap."

In other words, an email address is a pretty unique reference of a person.
Therefore, all I need to do is to give my mailbox a SW "homepage" rather
than me...

   <rdf:Description
        rdf:about="mailto:sean@mysterylights.com" [...] />

What about security? Well, anyone can mask an SMTP channel, just as anyone
can write an assertion about my mailbox. So, the easy and obvious fix is to
digitally stamp/sign/+ both my mailbox and its RDF assertion with the same
signature. We'll have to wait for the SW to come around until that step can
be completed (I'll forgo it for now).
Also, some of you may have noticed the small problem: by saying that these
properties are about mailto:sean@mysterylights.com, aren't we just saying
that these are the properties of my mailbox? Yes, we are...but I don't have
a URI that describes *me*, so it'll have to do for now I suppose... (better
ideas on a postcard to [...]). I don't think anyone will honestly mind if
we abuse the mailto: protocol for a better use...

O.K., the next step is to reify some simple properties to make assertions
about me (represented by my mailbox) (I'll use the me: prefix for this):-

   <rdf:Description
        rdf:about="mailto:sean@mysterylights.com"
        me:name="Sean B. Palmer"
        me:affiliation="Freelance/WAI IE"
        ms:seeAlso="http://infomesh.net/sbp/"
        me:contact="http://infomesh.net/sbp/#contact" />

Aside: Does anybody know of actual namespaces for these properties?
Anyway, that is just the first step. Next we could apply the ontological
layer... Let's talk about sex. If I am male therefore I cannot be female.
Let's add that property to the RDF:-

   <rdf:Description
        rdf:about="mailto:sean@mysterylights.com"
        me:name="Sean B. Palmer"
        me:affiliation="Freelance/WAI IE"
        ms:seeAlso="http://infomesh.net/sbp/"
        me:contact="http://infomesh.net/sbp/#contact"
        me:sex="m" />

If the "me:" namespace contained some DAML(?) to the effect that "if A is
male, then A is not female", that would be the first step in making this
little snippet of SW data logical and truly semantic. Anyway, you can see
where I'm going with this little example, so I'm going to bail out now
before it gets too complicated. What it shows you is:-
1. The SW doesn't have to be complicated for simple assertions
2. A URI makes a good namespace and foundation principle for the SW
3. Ontology and logic has it's uses

I hope this idea gets furthered a little bit, and developed...

Kindest Regards,
Sean B. Palmer
http://infomesh.net/sbp/
http://www.w3.org/WAI/ [ERT/GL/PF]
"Perhaps, but let's not get bogged down in semantics."
   - Homer J. Simpson, BABF07.