Could you please recommend for me the best heap size to reason about large
and expressive ontologies in my machine. In what follows my settings:
1/java -Xmx950M -Xms200M -Xss16M -XX: (is it ok? I changed the -Xmx500 to
-Xmx950M)(in the run.bat file)
You control the heap size available to the Java process running Protege
with the -Xmx parameter, just as you did.
950MB is probably not enough for reasoning over a "very large and
expressive" ontology. You may need several gigabytes. I see that your
system has 4 GB of memory, in total (which is not that much, really),
and you are using about 3GB already (I am not sure whether this was
while you were running Protege or before/after it). I would try to close
all other memory consuming applications, and start Protege with 2GB of
memory, and see if it can run the reasoner. However, it is hard to give
a definitive answer without knowing more about your ontology. What do
you mean by large and expressive ontology? How many axioms? What kind of
axioms? Maybe if you could share with us the content of the Ontology
Metrics view, after you loaded the ontology, it would give us a better idea.
First, I thank you very much for your help.
Those settings (sent) were used when I was running protege.
I changed the heap size to 2000 and 1500, but protege could not start. So I
set it to 1200 and I just closed applications consuming memory. the
performances of reasoning haven't changed.
Here are some of myontology metrics:
The second screen shot (with information about the more complicated
ontology) is unreadable. Can you please resend it?
I suppose that you can reason over the test ontology, right?
Which reasoner have you tried using? I'd suggest you try different
reasoners, as each reasoner has different performance.
If you could share the university ontology with us (even in private, if
you rather not share it on the list) we could make some experiments to
see how much memory is enough for running the reasoner. Or you can try
it yourself on a different machine, with more reasoners. Other that
that I have no further ideas how to help you.