[Ontology-editors] GO Ontology QA (fwd)

[Midori Harris]

fyi

---------- Forwarded message ----------
Date: Tue, 16 Jun 2009 17:01:19 -0600
From: Karin Verspoor <Karin.Verspoor at ucdenver.edu>
To: jdeegan at ebi.ac.uk, cjm at fruitfly.org, dph at informatics.jax.org,
     rhee at acoma.Stanford.EDU, midori at ebi.ac.uk, huala at acoma.Stanford.EDU,
     jblake at informatics.jax.org, musen at stanford.edu
Cc: Larry Hunter <Larry.Hunter at ucdenver.edu>
Subject: GO Ontology QA

This was just logged in the sourceforge tracker system.  It was
suggested to me by Larry Hunter that since you were participants in
the workshop that defined the Ontology QA Univocality problem, you
would be interested to see the analysis.

Thank you,
Karin



-------------

As a result of the analysis described in the ISMB paper http://bioinformatics.oxfordjournals.org/cgi/content/full/btp195?ijkey=qvYcpVnIJMjd19Y&keytype=ref
   "Ontology quality assurance through analysis of term
transformations" we have identified groups of terms in the GO that are
not "univocal", where some rephrasing is warranted.

Some general observations:

** in general, all "the" "an" and "a" determiners can be removed
** in general look for terms with "or" or "and" and eliminate the
coordination in favor of 2 specific terms (or eliminate completely --
in some cases it appears to be irrelevant)
** look at use of punctuation, in particular "," which in some cases
seems to be used in place of a specific relational preposition
**consider use of "within" vs "in"
** generally it seems that "other" or "another" is not necessary

The clusters that should be looked at are in the attached file.  I'd
appreciate knowing which of these turn out to be "true" true positives
that result in some change in phrasing of the GO terms.

Thanks,
Karin

-- 
Karin Verspoor, PhD
Research Assistant Professor
Center for Computational Pharmacology, University of Colorado Denver
PO Box 6511, MS 8303, Aurora, CO 80045 USA
karin.verspoor at ucdenver.edu / tel: (720) 279-4875 / campus: 4-3758