Ability of Protege to handle large ontologies

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

Ability of Protege to handle large ontologies

Michael DeBellis-2
I apologize I also posted this question on the ontolog list. Yesterday I was just assuming that the requirements of my client were too large for Protege and would require some kind of triple store product but this morning I'm realizing I shouldn't just assume that. Especially because the way they are going to use OWL, at least for the time being, won't involve a lot of real time updates or any support for concurrent users. 

So here's the question: could Protege, given a big enough machine and memory, handle an ontology with approximately 30K classes and instances? (Their model is unique due to their specific application requirements, there will be only a few more instances than classes). Could Protege handle 100K classes and instances? 

Also, I know nothing about triple store products. Someone on ontolog recommended blazegraph.com.  If anyone has suggestions for other products I would be interested. 

Michael

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

Re: Ability of Protege to handle large ontologies

Lorenz Buehmann

From my point of view, I don't think it scales that well - especially since most of the things in Java Swing (e.g. the class hierarchy tree) are not lazy and the UI might become unresponsive quite fast. But, I never tested it with more than 16G of memory, I guess the developers should know it better.


Triple stores: this is a very opinion-based question. Common triple stores are GraphDB, Virtuoso, Blazegraph, Jena TDB, etc.  And I'm a big fan of Stardog - it supports to what I know the most complex reasoning (yet, only via query rewriting). It's also dependent on the need for a free version or commercial one, and which kind of features besides SPARQL queries you need. For example, some also support full-text indexing which might be much faster than SPARQL REGEX.


Cheers,

Lorenz


On 30.08.2017 14:20, Michael DeBellis wrote:
I apologize I also posted this question on the ontolog list. Yesterday I was just assuming that the requirements of my client were too large for Protege and would require some kind of triple store product but this morning I'm realizing I shouldn't just assume that. Especially because the way they are going to use OWL, at least for the time being, won't involve a lot of real time updates or any support for concurrent users. 

So here's the question: could Protege, given a big enough machine and memory, handle an ontology with approximately 30K classes and instances? (Their model is unique due to their specific application requirements, there will be only a few more instances than classes). Could Protege handle 100K classes and instances? 

Also, I know nothing about triple store products. Someone on ontolog recommended blazegraph.com.  If anyone has suggestions for other products I would be interested. 

Michael


_______________________________________________
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: Ability of Protege to handle large ontologies

Balhoff, James Patrick
In reply to this post by Michael DeBellis-2
Here is an ontology with 129k classes (no instances) you can load: 


It works okay on my laptop; I have a MacBook Pro with 16 GB memory. I configure Protege to use up to 12 and I think this ontology requires around 6.

As Lorenz said the main problem is that the UI often becomes unresponsive and you need to wait a little while for it to become ready again. The other thing to consider is what reasoner you plan to use. With ELK you can classify that ontology in a few seconds, however if you include instances it can be a completely different story.

For a triple store I usually use Blazegraph. But typically with a lot of OWL reasoner preprocessing before loading.

Best regards,
Jim



On Aug 30, 2017, at 8:20 AM, Michael DeBellis <[hidden email]> wrote:

I apologize I also posted this question on the ontolog list. Yesterday I was just assuming that the requirements of my client were too large for Protege and would require some kind of triple store product but this morning I'm realizing I shouldn't just assume that. Especially because the way they are going to use OWL, at least for the time being, won't involve a lot of real time updates or any support for concurrent users. 

So here's the question: could Protege, given a big enough machine and memory, handle an ontology with approximately 30K classes and instances? (Their model is unique due to their specific application requirements, there will be only a few more instances than classes). Could Protege handle 100K classes and instances? 

Also, I know nothing about triple store products. Someone on ontolog recommended blazegraph.com.  If anyone has suggestions for other products I would be interested. 

Michael
_______________________________________________
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: Ability of Protege to handle large ontologies

Michael DeBellis-2
James and Lorenz, thanks that's great feedback. I think we may be able to make do with Protege after all.  Especially because there won't be a lot of use of defined classes or rules in the ontology. 

Michael

On Wed, Aug 30, 2017 at 7:38 AM, James Balhoff <[hidden email]> wrote:
Here is an ontology with 129k classes (no instances) you can load: 


It works okay on my laptop; I have a MacBook Pro with 16 GB memory. I configure Protege to use up to 12 and I think this ontology requires around 6.

As Lorenz said the main problem is that the UI often becomes unresponsive and you need to wait a little while for it to become ready again. The other thing to consider is what reasoner you plan to use. With ELK you can classify that ontology in a few seconds, however if you include instances it can be a completely different story.

For a triple store I usually use Blazegraph. But typically with a lot of OWL reasoner preprocessing before loading.

Best regards,
Jim



On Aug 30, 2017, at 8:20 AM, Michael DeBellis <[hidden email]> wrote:

I apologize I also posted this question on the ontolog list. Yesterday I was just assuming that the requirements of my client were too large for Protege and would require some kind of triple store product but this morning I'm realizing I shouldn't just assume that. Especially because the way they are going to use OWL, at least for the time being, won't involve a lot of real time updates or any support for concurrent users. 

So here's the question: could Protege, given a big enough machine and memory, handle an ontology with approximately 30K classes and instances? (Their model is unique due to their specific application requirements, there will be only a few more instances than classes). Could Protege handle 100K classes and instances? 

Also, I know nothing about triple store products. Someone on ontolog recommended blazegraph.com.  If anyone has suggestions for other products I would be interested. 

Michael
_______________________________________________
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



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