Re: [xtm-wg] name - occurrence? (food for thought) was: Topic Naming Constraint

[Robert Barta <rho@bigpond.net.au>]

> Nikita Ogievetsky wrote:
> 
> Some times topic's identity can be formed
> by a set of characteristics along with topic name.
> For example, for merging purposes my address
> could be as good as my name if not better.
> My date of birth is another candidate.
> 
> All three together are even better!

What I have understood from the various papers is, that there
are supposedly two ways to identify topics:

  (a) explicit via an external name space identifier. Good candidates
    are languages (finite, well), countries, ... maybe DNA samples
    for people (urn:dna:CTTGUTC..... :-)

  (b) implicit via 'some common attributes', be it <baseName>s or
whatever

(a) is decent and proper, (b) will haunt us forever because it silently
introduces a semantical level:

  - when are two names ought to be identical?
  - when two URLs?
  - what combination of attributes count? What, if there are collisions?
    what if it is ambiguous?

(b) is actually something which I NEVER plan to adopt into my
implementations. I rather leave it up to the map administration
to either merge the topics explicitely or let them be different.
'Silent merging' is not controllable from a resource-management
point of view. From my algebraic gutts I wonder about the mathematical
properties of the 'merging' operator.

<More arguments withhold/>

> Also when I see:
> <baseName>
>     <scope><topicRef xlink:href="#Dallas"/></scope>
>     <baseNameString>Steve Newcomb</baseNameString>
> </baseName>
> (Steve, forgive me I keep using your name from the example you
> suggested)
> 
...
> In fact, what happened here is that one of
> Steve's characteristics (address) becomes a scope of his name!
> I believe it is wrong!
> This is completely conceptually different from scoping
> a name with the language it is expressed in!

No, it is only wrong if you start to impose a application semantics
(not talking about map semantics). Once you abstract and see it as
a 'topic' only, above construct is perfectly OK.

> I claim that my address is not the scope of my name but my occurrence!
> And that my name is just another <resourceData> occurrence.
> And <subjectIdentity> can reference all these occurrences to identify
> an individual.
> So if name is not enough to identify the subject, - add address and
> date of birth
> and make it an "AND".

Again, ____PLEASE_____ no more implicit boolean expressions. They only 
make a mess for a (formal) model, introduce asymmetries, ..., world
famine....

\rho

To Post a message, send it to:   xtm-wg@eGroups.com

To Unsubscribe, send a blank message to: xtm-wg-unsubscribe@eGroups.com