| RE: [higgins-dev] Data Model: Many same-typed attributes? |
|
Let’s definitely make the Higgins wiki
pages under http://wiki.eclipse.org/Higgins_Data_Model
authoritative. AND in the process let’s document and work out any differences
with HOWL. (As a side benefit of doing that, we’ll also be able to document any
differences with the XDI RDF model so that Higgins deployments can can cleanly send
data in the HDM around in XDI. A triple win!) =Drummond From:
higgins-dev-bounces@xxxxxxxxxxx [mailto:higgins-dev-bounces@xxxxxxxxxxx] On Behalf Of Jim Sermersheim This
is why I started wondering about an "authoritative" data model.
If we say the wiki pages are authoritateve, it allows us to define the
model exactly the way we want to. The down side is that we might design
something (like this) which can't be represented in the HOWL. OTOH,
making the HOWL authoritative might cause the APIs to become clunky. From: higgins-dev-bounces@xxxxxxxxxxx
[mailto:higgins-dev-bounces@xxxxxxxxxxx] On Behalf Of Jim
Sermersheim Sent: Monday,
January 14, 2008 1:38 PM To: Higgins
dev <higgins-dev@xxxxxxxxxxx Subject: [higgins-dev]
Data Model: Many same-typed attributes? Mike also
resurrected this topic in http://dev.eclipse.org/mhonarc/lists/higgins-dev/msg03734.html
Currently
we say there are 0..1 instances of a given attribute on a subject, and for any
attribute, there are 1..N values. This question
is: Should we change it so we allow 0..N instances of a given
attribute, each with 1 value. In the
past, this was brought up because it would better align the IdAS APIs with the
way OWL or HOWL works. From an
API point of view, I prefer the grouping one gets with a single attribute with
multiple values. I don't
recall other arguments either way. If no one is interested in
re-hashing this topic, I'll let it die. Jim [=Drummond] Jim, I’m not so much interested in re-hashing this
topic…as getting to the bottom of it. This is one of those fundamental
ontology/data model design decisions that then resonates throughout the entire
model. From a newcomer’s POV, I think it’s fascinating that the RDF/OWL model
is 0..N instances of an attribute, each with 1 value, while the Higgins Data
Model is 0..1 instances of an attribute with 1..N values. Whatever the ultimate decision is, this is one topic that deserves
a really crisp explanation in the upgraded documentation for the Higgins Data
Model. Again, I’m willing to help, but this probably warrants a discussion at
the F2F first – unless you think we can seriously advance it via
email/wiki between now and then. =Drummond |