Natasha, Peter, et al.
I test the COSMO using vowlidator; Protégé 3.4beta (build 533 can't
handle rdfs:class for some reason); Protégé 4 (Fact++ reasoner); and SWOOP,
and run the consistency checks. Even then there may be errors that are not
directly detected, which show up only after I have saved the ontology in
Protégé 3.4, reloaded the saved ontology, and then check for the statistics:
At that point I sometimes get error messages that refer to some default
class or another, without a hint as to where the real problem is. I have to
resort to deleting or commenting out segments of the ontology to find the
source of the error, which often resides in the erroneous use of rdfs:class
for owl:class or vice-versa.
I create problems for myself by hand-editing the OWL file, because when
Protégé saves, it completely scrambles the order of the elements, loses the
comments, and nests the entities so deeply that it is hard to find the
logical description and documentation for individual classes in the OWL
file. At the current point I am still importing OWL classes as text, and
find it best to control the structure of the source file regardless of the
nuisance when obscure errors are reported. Likewise I prefer to use
rdfs:Class rather than owl:Class for metatypes, it makes the intended
meanings clearer to inspection.
I looked at the output log, and couldn't find any information to indicate
the source of the error in r875.
There was one version of COSMO rev875 that did have one of those obscure
errors (before testing), which was then removed. I do not know if that was
the version uploaded, or whether it was a clean tested version. I have just
uploaded COSMOrev877, which went through all the tests I mentioned. We
shall see if it parses properly. It would really be nice if those error
reports could mention the name of the entity that is causing the problem.
If it goes OK, then there may have been one of those obscurely reported
errors in r875. If not, I would need some way of parsing segments of the
ontology to try to isolate the problem as perceived by the oor parser. (01)
> -----Original Message-----
> From: oor-dev-bounces@xxxxxxxxxxxxxxxx [mailto:oor-dev-
> bounces@xxxxxxxxxxxxxxxx] On Behalf Of Natasha Noy
> Sent: Wednesday, February 17, 2010 2:22 PM
> To: OpenOntologyRepository-development
> Subject: Re: [oor-dev] Parsing Ontologies [was - Fwd: registration...
> oor sandbox (v2.x)]
> A good test is usually to see if the ontology opens up fine in Protege
> 3.4. That's the version of Protege that BioPortal uses. If Protege
> can't handle it, BioPortal won't be able to do it either.
> On Feb 17, 2010, at 11:15 AM, Patrick Cassidy wrote:
> > Paul,
> > Thanks for looking into this. The COSMO has over 7000 classes,
> > over 700
> > relations and about 1600 restrictions, and is multi-inheritance. So
> > it is
> > more complex than many domain or utility ontologies. It also uses
> > rdfs:class as well as owl:class, which seems to confuse at least one
> > version
> > of Protégé (though not others!).
> > If the parser throws off some error message I'd like to see it.
> > Pat
> > Patrick Cassidy
> > MICRA, Inc.
> > 908-561-3416
> > cell: 908-565-4053
> > cassidy@xxxxxxxxx
> >> -----Original Message-----
> >> From: Paul R Alexander [mailto:palexander@xxxxxxxxxxxx]
> >> Sent: Wednesday, February 17, 2010 12:56 PM
> >> To: Peter Yim
> >> Cc: OpenOntologyRepository-development; Patrick Cassidy; Tejas
> >> Subject: Re: Parsing Ontologies [was - Fwd: registration... oor
> >> sandbox
> >> (v2.x)]
> >> Peter,
> >> It sounds like there was a failure at some point during the ontology
> >> parsing process, meaning that the process had started, setting the
> >> ontology status to "Parsing", but never finished, so the ontology
> >> status
> >> was never updated. There is a "Parsing Error" status, I'm not sure
> >> why
> >> that wasn't set if a problem was encountered. I would try to login
> >> the BioPortal Admin software and trigger a re-parse on Pat's
> >> ontology.
> >> If it does encounter an error during the process, the Admin UI
> >> report that back to you.
> >> Paul
> >> On 2/17/10 8:41 AM, Peter Yim wrote:
> >>> Paul or Someone from NCBO who might point us in the right
> >> direction ...
> >>> Further to the earlier exchange with PatCassidy (below), looks like
> >>> his COSMO ontology (which Pat uploaded on 2010.02.11) is still
> >>> "being
> >>> parsed" at the OOR-Sandbox (v2.x).
> >>> (Not knowing what exactly to look for) I just went into the server
> >> and
> >>> check cpu workload on the<oor-01> vm. To my surprise, the cpu(s)
> >>> running at 99.5% to 100% idle. ... I do notice that 19 (of the 116)
> >>> ontologies that are listed in the "Browse" page are supposed to be
> >>> in
> >>> the state of "Parse Ontology." ... therefore, this doesn't sound
> >>> right!
> >>> ... Suggestions?
> >>> Thanks& regards. =ppy
> >>> --
> >>> ---------- Forwarded message ----------
> >>> From: Patrick Cassidy<pat@xxxxxxxxx>
> >>> Date: Fri, Feb 12, 2010 at 7:31 PM
> >>> Subject: RE: registration problem oor sandbox (v2.x)
> >>> To: Peter Yim<peter.yim@xxxxxxxx>
> >>> Cc: Tejas Parikh<tejas@xxxxxxxxxxxxxxx>, Paul R Alexander
> >>> <palexander@xxxxxxxxxxxx>, Ken Baclawski<kenb@xxxxxxxxxxx>, "Obrst,
> >>> Leo J."<lobrst@xxxxxxxxx>, Mike Dean<mdean@xxxxxxx>
> >>> On the parsing time for the new sandbox:
> >>>>> [PatC] Any idea how long a 7000 element ontology should take to
> >>>> parse?
> >>>> [ppy] (I'm guessing, but) I don't think anyone knows* ... Please
> >> make
> >>>> a note on how long it is taking this time for COSMO, maybe when
> >>>> are done, you can share the experience with all of us. [ *since
> >>>> different installs are running on fairly different hardware and
> >>>> platforms, and experiencing different work loads all the time. ]
> >>> The COSMO is still "parsing" after over 24 hours. I see a parsing
> >> error for
> >>> the OWL-time ontology (didn't see that in the last version). Who
> >>> maintaining the parser and display function? Any thoughts on how
> >> this can
> >>> be improved?
> >>> The "explore" function is really valuable. I hope we can get it to
> >> be
> >>> consistently effective. Sorry I have no programming expertise to
> >> help out
> >>> here.
> >>> Pat
> >>> Patrick Cassidy
> >>> MICRA, Inc.
> >>> 908-561-3416
> >>> cell: 908-565-4053
> >>> cassidy@xxxxxxxxx
> > _________________________________________________________________
> > Message Archives: http://ontolog.cim3.net/forum/oor-dev/
> > Config/Unsubscribe: http://ontolog.cim3.net/mailman/listinfo/oor-dev/
> > Shared Files: http://ontolog.cim3.net/file/work/OOR/dev/
> > Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OpenOntologyRepository
> Message Archives: http://ontolog.cim3.net/forum/oor-dev/
> Config/Unsubscribe: http://ontolog.cim3.net/mailman/listinfo/oor-dev/
> Shared Files: http://ontolog.cim3.net/file/work/OOR/dev/
> Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OpenOntologyRepository (04)
Message Archives: http://ontolog.cim3.net/forum/oor-dev/
Shared Files: http://ontolog.cim3.net/file/work/OOR/dev/
Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OpenOntologyRepository (05)