[protege-discussion] Working with large ontologies in Web Protege 2

[Timothy Redmond]

On 9/9/12 5:38 PM, Rajeev Singh wrote:
> Thank you so much for the response.and clarifying that Web Protege 2 
> does not send the entire ontology to the client
>
> Good to know that 90000 class ontology can be loaded in Protege 4 in 
> 15 secs. Please tell me more about the machine configuration which 
> gives this kind of performance.

I ran this on a mac with one 2.66 Ghz Intel core i5 processor (4 
cores).  I gave Protege a lot of memory but my recollection is that the 
performance is still good down to something like 800M.

> Going with the assumption that after loading a large ontology the 
> response time of operations does not become sluggish.
The response time should not become sluggish.  In memory ontologies 
usually give quite good performance.  Sluggishness comes from using a 
database backend or doing some heavy-weight processing such as running a 
reasoner.


-Timothy


>
> Also wanted to clarify if web Protege 2 (OWL API based) can use a DB 
> back end or it just works with file based storage.
>
> On Thu, Sep 6, 2012 at 11:27 AM, Timothy Redmond 
> <tredmond at stanford.edu <mailto:tredmond at stanford.edu>> wrote:
>
>
>     I am not the best person to answer this but I can give some
>     answers and forward it to people who will give more information.
>
>
>     On 09/05/2012 04:09 PM, Rajeev Singh wrote:
>>     Hi,
>>     I aplogize for re-sending. Going with the assumption that my
>>     earlier mail (08/31) fell through the cracks.
>>     The features in WebProtege 2 are very promising and will
>>     definitely help in increasing adoption of semantic technologies
>>     in the area of knowledge
>>     management. Thanks for this valuable effort.
>>     Have a couple of questions regarding working with large
>>     ontologies in WebProtege. Let us assume that we are working with
>>     ontology with about
>>     50000 TBox & 100000 ABox nodes. Also assume that each TBox class
>>     has 10 properties(5 data and 5 object)
>
>     50,000 classes is not really very large.  Protege 4 loads
>     ontologies with about 90,000 classes in less than 15 seconds.  In
>     the Protege 3 Web-Protege, putting this information into a
>     database will take some time but it should still be very workable.
>
>     I have less experience with large numbers of individuals.
>
>
>>     - How does WebProtege store ontologies. As RDF files or in RDF
>>     triple store (DB based)
>
>     Protege 3 based WebProtege stores the ontology in a Protege 3
>     database.  It also can be configured to load the ontology in
>     memory which would run a lot faster but which requires a server
>     with more memory.
>
>     The OWL api based Web Protege stores the ontology on disk and in
>     memory.  It uses a file format that allows it to load large
>     ontologies very quickly (SNOMED in under 2 seconds) so it can be
>     somewhat aggressive about removing an in-memory ontology.
>
>
>>     - Given a massive ontology does the back end server load the
>>     entire ontology in memory when a web Client opens it
>
>     Protege 3 based web protege can store the ontology in a Protege 3
>     database.
>
>
>>     - When a web client opens a large ontology is the entire ontology
>>     sent to the client
>
>     Web Protege never sends the entire ontology to the client.
>
>     -Timothy
>
>
>>
>>     Thanks, -Rajeev
>>
>>
>>     _______________________________________________
>>     protege-discussion mailing list
>>     protege-discussion at lists.stanford.edu  <mailto:protege-discussion at lists.stanford.edu>
>>     https://mailman.stanford.edu/mailman/listinfo/protege-discussion
>>
>>     Instructions for unsubscribing:http://protege.stanford.edu/doc/faq.html#01a.03
>
>
>     _______________________________________________
>     protege-discussion mailing list
>     protege-discussion at lists.stanford.edu
>     <mailto:protege-discussion at lists.stanford.edu>
>     https://mailman.stanford.edu/mailman/listinfo/protege-discussion
>
>     Instructions for unsubscribing:
>     http://protege.stanford.edu/doc/faq.html#01a.03
>
>
>
>
> _______________________________________________
> protege-discussion mailing list
> protege-discussion at lists.stanford.edu
> https://mailman.stanford.edu/mailman/listinfo/protege-discussion
>
> Instructions for unsubscribing: http://protege.stanford.edu/doc/faq.html#01a.03

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.stanford.edu/pipermail/protege-discussion/attachments/20120910/bd41db8f/attachment-0001.html>