[MOBY-dev] [moby] Re: Cleaning up the Object ontology - Inheriting from base Object

[Mark Wilkinson]

On Fri, 2006-02-17 at 09:28 -0700, Paul Gordon wrote:

> object without dying a horrible death.  I think proper namespace 
> restriction in service registration is a bigger issue than people 
> creating new name-only ontology objects...


I agree, and I believe that much of this confusion is happening because:

1)  My fault - insufficient documentation of what the namespace *means*
in MOBY.  It's discussed in publications and such, but not discussed
clearly in the tutorials.

2)  My fault - the Namespaces have been modelled incorrectly (IMO) right
from the start as a flat list, because of our need (or at least desire)
to be compatible with the GO community.  I think this has to change - we
either have to break away from them again, or better, bring them with
us.  Either way, we need to allow for hierarchical namespaces such that
people do not need to mix the semantics of what an object represents
(the Namespace) with the syntax of how it is represented (the Object)

The short-term gain that people are observing by creating domain-
specific Objects is just that - short term gain.  It is not an
extensible solution, since it leads to a combinatorial explosion of
data-types:

Object
   DatabaseObject
       GenbankObject
          GenbankSequence
             GenbankDNASequence
             GenbankAASequence
       EMBLObject
          EMBLSequence
             EMBLDNASequence
             EMBLAASequence
       PDBObject
          PDBSequence

etc. etc.  From the Rectorian ontology-philosophy, this is a perfect
example of the "exploding bicycle", and the end result is well
documented and disasterous!  Even in that example, there would be no way
to discover a Blast service that operated on GenericSequence objects if
you had a GenbankDNASequence object in-hand.

I **desperately** want to solve this problem.  The discussion over the
past couple of days has revealed that there is a serious need for a
rapid solution since several of our key users are already starting down
this path.  

I maintain that the solution is NOT to generate domain-specific objects,
but rather to model the orthogonal Namespace ontology properly, and
facilitate hierarchical searching over Namespaces in the registry, such
that creating these mixed objects is no longer necessary.  If I am
understanding correctly the reason why people are creating these
objects, then this srikes me as being the Right Thing to do.

Am I mis-interpreting the problem?  If so, please explain it to me.

M


-- 
--
...his last words were 'Hey guys!  Watch this!'
--
Mark Wilkinson
Asst. Professor
Dept. of Medical Genetics
University of British Columbia
PI in Bioinformatics
iCAPTURE Centre
St. Paul's Hospital
Rm. 166, 1081 Burrard St.
Vancouver, BC, V6Z 1Y6
tel: 604 682 2344 x62129
fax: 604 806 9274

"For most of this century we have viewed communications as a conduit, 
       a pipe between physical locations on the planet. 
What's happened now is that the conduit has become so big and interesting 
      that communication has become more than a conduit, 
       it has become a destination in its own right..."

                Paul Saffo - Director, Institute for the Future