Pellet is unresponsive

classic Classic list List threaded Threaded
4 messages Options
Reply | Threaded
Open this post in threaded view
|

Pellet is unresponsive

Ehsan Abdolmajidi
Hello everyone,

I have been working with ontology and RDF using Protege as the
developing environment for a while. I developed my ontology and am now
ready to reason over it. Running Hermit reasoner returns the inferences
but Pellet seems cannot finish the inference. It is stuck in the "same
individual" step. My ontology has many distinct instances and I know it
loads a lot of computation time but Hermit is fast. Moreover, I had the
second ontology almost as big as the other one which is being inferred
by Pellet in ~30 min. Anyhow, I cannot figure out what is wrong with my
ontology and how I can fix it as no error is logged either. I increased
the heap size to 10G but still not working.
I must also mention that I need pellet because (to my knowledge) Hermit
cannot reason the SWRL rules which involve "Data Property" relations.

I appreciate if you can help me with the issue. I attached my ontologies
here: NVDB.owl is working with Pellet but OSM.owl is not. Please don't
hesitate to contact me for further info.

Best regards,
Ehsan

--
Ehsan Abdolmajidi
PhD Student
GIS center
Department of Physical Geography and Ecosystem Science
Lund University, Sweden
http://www.gis.lu.se/english/contact.htm


_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user

NDVB.owl (359K) Download Attachment
OSM.owl (700K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Pellet is unresponsive

Tania Tudorache
Hi Ehsan,

This is not a Protege question. I suggest to repost it on the Pellet mailing list:
https://groups.google.com/forum/?fromgroups#!forum/pellet-users

Probably some constructs in your ontology make the reasoning take so long with Pellet. I am sure you will get a good answer on the Pellet mailing list, please post there.

Cheers,
Tania



On 04/25/2016 03:33 AM, Ehsan Abdolmajidi wrote:
Hello everyone,

I have been working with ontology and RDF using Protege as the developing environment for a while. I developed my ontology and am now ready to reason over it. Running Hermit reasoner returns the inferences but Pellet seems cannot finish the inference. It is stuck in the "same individual" step. My ontology has many distinct instances and I know it loads a lot of computation time but Hermit is fast. Moreover, I had the second ontology almost as big as the other one which is being inferred by Pellet in ~30 min. Anyhow, I cannot figure out what is wrong with my ontology and how I can fix it as no error is logged either. I increased the heap size to 10G but still not working.
I must also mention that I need pellet because (to my knowledge) Hermit cannot reason the SWRL rules which involve "Data Property" relations.

I appreciate if you can help me with the issue. I attached my ontologies here: NVDB.owl is working with Pellet but OSM.owl is not. Please don't hesitate to contact me for further info.

Best regards,
Ehsan



_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user


_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user
Reply | Threaded
Open this post in threaded view
|

Re: Pellet is unresponsive

Ehsan Abdolmajidi
Hi Tania,

Thanks for your reply.
This problem seems to be encountered before and there was a relevant question in the Pellet mailing list. Apparently, the recursion in JAVA is not efficient and causes this problem. I am hoping they have found a solution but haven't yet heard from them.

Meantime, I am wondering if you can kindly help me find the responsible construct for such a problem. How could I detect which part of my ontology is troubling?

Thank you for your time and appreciate any help.

Regards,
Ehsan

On 4/26/2016 12:32 AM, Tania Tudorache wrote:
Hi Ehsan,

This is not a Protege question. I suggest to repost it on the Pellet mailing list:
https://groups.google.com/forum/?fromgroups#!forum/pellet-users

Probably some constructs in your ontology make the reasoning take so long with Pellet. I am sure you will get a good answer on the Pellet mailing list, please post there.

Cheers,
Tania



On 04/25/2016 03:33 AM, Ehsan Abdolmajidi wrote:
Hello everyone,

I have been working with ontology and RDF using Protege as the developing environment for a while. I developed my ontology and am now ready to reason over it. Running Hermit reasoner returns the inferences but Pellet seems cannot finish the inference. It is stuck in the "same individual" step. My ontology has many distinct instances and I know it loads a lot of computation time but Hermit is fast. Moreover, I had the second ontology almost as big as the other one which is being inferred by Pellet in ~30 min. Anyhow, I cannot figure out what is wrong with my ontology and how I can fix it as no error is logged either. I increased the heap size to 10G but still not working.
I must also mention that I need pellet because (to my knowledge) Hermit cannot reason the SWRL rules which involve "Data Property" relations.

I appreciate if you can help me with the issue. I attached my ontologies here: NVDB.owl is working with Pellet but OSM.owl is not. Please don't hesitate to contact me for further info.

Best regards,
Ehsan



_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user



_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user

-- 
Ehsan Abdolmajidi
PhD Student
GIS center
Department of Physical Geography and Ecosystem Science
Lund University, Sweden
http://www.gis.lu.se/english/contact.htm

_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user
Reply | Threaded
Open this post in threaded view
|

Re: Pellet is unresponsive

Rafael Gonçalves
Hi Ehsan,

I’m afraid that what you ask for is not so simple, in particular pinpointing specific OWL construct(s) without serious experimentation. Since you mentioned that the reasoner is stuck computing 'same individuals', have you tried changing the reasoner preferences so that it doesn’t compute those kinds of entailments?

There are methods for identifying subsets of an ontology that are performance-degrading for a given ontology-reasoner pair [1], and an associated prototype tool [2]. These subsets are considered performance “hot spots” if their removal from the ontology gives rise to a significant performance improvement. That doesn’t necessarily mean that you’ll find “the expensive construct(s)” in the hot spot (if there ever were any in particular to begin with), indeed you may find no suspicious constructs whatsoever, but the bottom line is that the interaction between axioms in the hot spot and axioms in the remaining ontology is giving the specified reasoner a hard time. By weakening or removing hot spot axioms, you should expect better performance overall. (The tool I mentioned can provide you with some approximations of the hot spot.)

Hope this helps.

Cheers,
Rafael


On Apr 26, 2016, at 09:21, Ehsan Abdolmajidi <[hidden email]> wrote:

Hi Tania,

Thanks for your reply.
This problem seems to be encountered before and there was a relevant question in the Pellet mailing list. Apparently, the recursion in JAVA is not efficient and causes this problem. I am hoping they have found a solution but haven't yet heard from them.

Meantime, I am wondering if you can kindly help me find the responsible construct for such a problem. How could I detect which part of my ontology is troubling?

Thank you for your time and appreciate any help.

Regards,
Ehsan

On 4/26/2016 12:32 AM, Tania Tudorache wrote:
Hi Ehsan,

This is not a Protege question. I suggest to repost it on the Pellet mailing list:
https://groups.google.com/forum/?fromgroups#!forum/pellet-users

Probably some constructs in your ontology make the reasoning take so long with Pellet. I am sure you will get a good answer on the Pellet mailing list, please post there.

Cheers,
Tania



On 04/25/2016 03:33 AM, Ehsan Abdolmajidi wrote:
Hello everyone,

I have been working with ontology and RDF using Protege as the developing environment for a while. I developed my ontology and am now ready to reason over it. Running Hermit reasoner returns the inferences but Pellet seems cannot finish the inference. It is stuck in the "same individual" step. My ontology has many distinct instances and I know it loads a lot of computation time but Hermit is fast. Moreover, I had the second ontology almost as big as the other one which is being inferred by Pellet in ~30 min. Anyhow, I cannot figure out what is wrong with my ontology and how I can fix it as no error is logged either. I increased the heap size to 10G but still not working.
I must also mention that I need pellet because (to my knowledge) Hermit cannot reason the SWRL rules which involve "Data Property" relations.

I appreciate if you can help me with the issue. I attached my ontologies here: NVDB.owl is working with Pellet but OSM.owl is not. Please don't hesitate to contact me for further info.

Best regards,
Ehsan



_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user



_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user

-- 
Ehsan Abdolmajidi
PhD Student
GIS center
Department of Physical Geography and Ecosystem Science
Lund University, Sweden
http://www.gis.lu.se/english/contact.htm
_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user


_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user