Code Relationship

Renato, this is not something we have discussed in detail.  The closest
thing is that there is a possible genericode 3.0 requirement to support
multiple codes in a list for the same item.

Personally, I have two comments adding relationships like this between
codes:

* if you start adding relationships between codes, where do you stop?  If
you allow arbitrary relationships between codes, are you starting to
re-invent RDF/OWL?  I'm not suggesting that this requirement is
unimportant, or anything like that, only that it might require a different
standard, and there is a question as to what that standard would be (since
once people start building RDF/OWL relationships between things, it can be
hard to set boundaries on what they should or shouldn't do).  [I know
you've been in the RDF game a long time Renato, so by all means tell us
what you would do.]

* the current plan is for version 2.0 of genericode to add support for
generation of new code lists based on a set of input code lists and a set
of derivation rules.  Things like adding associations between codes, which
are straightforwards in a purely static code list model, are tricky when
you have derived code lists, because you need to be able to specify how
such associations are added, removed, or modified.  For this reason, if we
do consider adding support for such relationships, I think we need to put
it off to genericode 3.0.

In the short term (genericode 1.0), anyone who needs this information can
create a complex-valued (XML-valued) column in their code list, where the
XML contains a list of suitably related codes.  This is currently the only
way to introduce multi-valued data (i.e. where a single value in a row can
be replaced by an array of values) into genericode code lists, although
that is also a possible future requirement, and a possible way to
represent some kinds of associations.

Cheers, Tony.

On Mon, 23 Apr 2007 04:42:11 +0100, Renato Iannella