[MOBY-l] Which machine deals with ontology issues?

[Phillip Lord]

>>>>> "Mark" == mwilkinson  <mwilkinson at gene.pbi.nrc.ca> writes:

  Mark> Hi all,

  Mark> Now that there are a handful of services up and running, the
  Mark> issues around dealing with our wonderful Object/Service
  Mark> ontologies are becoming clearer to me.  I'm wondering if
  Mark> anyone has opinions about whether ontologies should be a
  Mark> Client/Service issue, or if they should be dealt with at
  Mark> MOBY-Central?

  Mark> e.g. I query MOBY central with an input Object-type.  Does it
  Mark> respond only with the services that deal with that object
  Mark> type, or does it track down the parentage of that Object and
  Mark> tell me the services that deal with that object type and its
  Mark> parents?  Similarly, does a Service accept *any* type of
  Mark> object, and skim-off the information that it understands based
  Mark> on the ontology, or does it restrict itself to only dealing
  Mark> with *exactly* the type of object that it is registererd to
  Mark> deal with, and reject everything else?

  Mark> I think these questions arise whether or not we have a
  Mark> separate ontology-server (i.e. rip ontology-awareness out of
  Mark> the MOBY-Central itself - which seems like a good idea in any
  Mark> case), as it remains ambiguous who has the responsibility for
  Mark> dealing with the ontologies - server, client, or Central?

  Mark> How does myGrid deal with this? 

We argue about it, and wonder which is best!

At the moment its an open issue for us. We are worried about the
amount of traffic that will go between the server and an ontology
driven query interface to the server. 

However there is no a priori reason, other than performance, why
ontological querying should be integrated into the server, so I
suspect that initially we will separate them, in the knowledge that
tighter integration may be necessary later. If this happens the
interfaces will probably still remain separate. 

Premature optimisation is after all the root of all evil.


Still we are open to suggestions, and this is something we should
probably talk about at ISMB/BOSC....

Phil