W3C home > Mailing lists > Public > public-owl-comments@w3.org > January 2009

haskey clarification

From: Marijke Keet <keet@inf.unibz.it>
Date: Thu, 29 Jan 2009 12:34:56 +0100
Message-ID: <49819460.1060200@inf.unibz.it>
To: public-owl-comments@w3.org

Dear Working Group members,

I'm a ware it is slightly after the deadline, so therefore I like to log 
only a clarfication on Haskey in addition to the one by Jim Hendler [1], 
which perhaps still could be added in the explanation.

 From conceptual data modelling and software engineering (w.r.t. 
identification) and ontology (w.r.t. identity), a key means a 1:1 
relation from the concept to its attribute(s), i.e., both functional and 
inverse functional, but not, as is in the current OWL2 draft, only 
inversefunctional (1:n).

Very briefly, and taking the draft's example, with the SSN one can 
indeed retrieve a unique instance of person, but not identify that 
person, as one person can have more than one SSN in this setting. With 
the latter, let's assume that, say, the Italian government counts its 
Italian citizens by their SSN (instead of lining them up in rows and 
physically counting the living physical appearances of these humans), 
then it may count more citizens in their information system than the 
actual amount of italians that live, with corresponding consequences, 
such as a person paying tax twice or receiving unemplyment benefits 
several times (as those services are normally linked to the SSN). To 
prevent this, one also has to add that each such citizen has at most 1 
SSN, so that there is a 1:1 correspondence and that the SSN indeed can 
be used as a surrogate for identifying persons (as approximation of the 
elusive identity criterion for person that Ontology tries to find out) 
and with which we thus reliably can count the population because each 
person is identified with that attribute.

Clearly, nothing prevents you from keeping the current Hashey 
terminology and semantics, but it may have some effect later on with 
modellers and implementers that give a different meaning to key.

[1] http://markmail.org/message/wgc5vgp5so5lmqne

Best regards,
Marijke

C. Maria Keet
Free University of Bozen-Bolzano
web: http://www.inf.unibz.it/krdb/
home: http://www.meteck.org <http://www.meteck.org/>
Received on Thursday, 29 January 2009 11:35:38 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 29 January 2009 11:35:39 GMT