Folks,
Nicolas raises an important point about ontology modularity. There are
two viewpoints one can take about modularity and both are important. One,
from the standpoint of modularity being needed to aid human understanding
or a large product. Two, from the standpoint of modularity being needed in
order to make computer implementation efficient from a space or speed
standpoint.
It should be possible for a human to learn only the portions of an
ontology that are needed for a particular application. Dependencies should
be clear, and modules that don't depend on each other should be
separate. That is the case in SUMO, which has 11 separable modules and a
documented dependency structure. The roughly two dozen different domain
ontologies and topics which extend SUMO are also provided in separate
files. More work does need to be done to document the dependencies of the
domain ontologies however.
With regard to implementation we should look at the model of modern
programming languages such as Java, which provide extensive libraries, but
only compile into binary form those which are actually used by a particular
program. I've taken the same approach in implementing formal ontology
applications. While extensive ontology definitions are available, only
what is actually used gets put into the running application. The
availability of resources does not necessarily slow or increase the size of
a running application which uses those resources. (01)
Adam
----------------------------
Adam Pease
http://www.ontologyportal.org - Free ontologies and tools (02)
_________________________________________________________________
Msg Archives: http://ontolog.cim3.net/forum/health-ont/
Community Files: http://ontolog.cim3.net/file/work/health-ont/NHIN-RFI/
To Post: mailto:health-ont@xxxxxxxxxxxxxxxx
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?NhinRfi (03)
|