Search Mailing List Archives
[protege-discussion] Saving inferred ontology
tredmond at stanford.edu
Fri Sep 27 14:29:17 PDT 2013
On 09/26/2013 04:49 PM, André Luiz Tietböhl Ramos wrote:
> Hello everyone,
> I have a relatively large ontology that takes a long time to be
> inferred. It is under development right now. In addition, I'd like to
> save the final inferred versions directly. Is it possible?
Protege 4 will let you export inferred axioms into an ontology. Use
File->Export Inferred Axioms as ontology.
> BTW, the best way to speed up the inference process to add more memory?
As far as I know, adding memory won't help speed things up for most
reasoners. I think that adding memory will only help, and in this case
it will help quite a bit, if you are right at the edge of running out of
memory for your java virtual machine. What happens in this case is that
sometimes java will run extremely slowly because it is spending
tremendous amounts of time doing garbage collection. But if java has
sufficient memory, adding more won't help.
Unfortunately I don't know of a general way of telling if the java
process lost in garbage collection. If you use the jdk version of java,
you can connect to the java program with jvisualvm and use this to
monitor the amount of garbage collection.
One other distinction is whether you are finding inference slow in that
period when it is initializing or if Protege 4 just runs too slowly
after the reasoner has initialized and you are moving about the
interface. If the initialization phase is too slow then your main
viable option is to try other reasoners.
If the Protege 4 user interface is too slow after the reasoner has
initialized, then this is happening because certain inferences that
Protege is displaying are taking too long to calculate. In this case,
you can control the inferences that Protege displays by configuring the
options in the Reasoner->Configure dialog. There is a youtube video
that tells how to do this:
This can also be used to control what the reasoner does when it
initializes but this is not usually recommended because it may result in
a slow-down of Protege after the reasoner has initialized because it has
not cached a sufficient set of inferences.
> Andre Luiz Tietbohl Ramos
> http://www.feng.pucrs.br/~andreltr <http://www.feng.pucrs.br/%7Eandreltr>
> protege-discussion mailing list
> protege-discussion at lists.stanford.edu
> Instructions for unsubscribing: http://protege.stanford.edu/doc/faq.html#01a.03
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the protege-discussion