Behavior of non-read-only imports

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
|

Behavior of non-read-only imports

Mark Feblowitz
I've been working heavily with a number of layered, imported, locally
stored ontologies, opting for the relatively safe but inconvenient
method of importing read-only and editing the various ontologies in
separate invocations of Protege-Owl.

I have noticed that the metadata tab supports the import of local
filesystem ontologies as read-only or writable.

I did have a stretch of "bad luck" recently when I used filesystem
URIs and found that multiple ontology files were altered (and
sometimes erased) due to changes when editing the "main"  ontology.
(I reported this bug but have heard nothing about it.) For this
reason, I switched to the much more conservative method of separately
editing files and importing only as read-only.

I have not yet found documentation on the non-read-only import
capability, nor do I fully understand the implications and
limitations of this capability. It appears that one can import
several ontology files and use the Metadata tab to select from among
the imported ontologies which one is currently active. From this I
infer that I can use the same invocation of Protege-Owl to edit
multiple ontologies, without the inconvenience of having to reload
the dependent ontologies after a change.

I assume that only a single ontology file can be active at any given
time. What specifically does this mean? I have noticed that, when I
make changes to the "main" ontology and save it, all of the other
ontologies that are open are also saved. I'm not sure why this is. It
certainly is a problem when an inconsistency arises, since empty
ontologies are saved out to some/all of the ontologies that were
imported as writable.

As far as the multi-ontology edits are concerned - is it possible to
"move" a definition from one ontology to another? There are times
when a concept needs to be "promoted" to another ontology (possibly a
shared ontology at a higher "layer"). This can be a very involved
operation, requiring the Class and (some of) the associated property
definitions to be moved to another namespace, and the migration of
all of these to "live" in another file. The former seems to be
covered the Refactor tool. But it appears that the latter needs to
occur via text editing. Is there a means of re-labeling the
definitions so that they'll be serialized out to a different ontology file?

Thanks,

Mark Feblowitz

-------------------------------------------------------------------------
To unsubscribe go to http://protege.stanford.edu/community/subscribe.html