Jena Pellet Performance

[rasha fawzy <ra...@gmail.com>]

Hi,

i tried to load a file (20 588 triple) using jena tdb and to reason this
file by pellet but the time the reasoning takes is unacceptable. To return
the results of a simple query, it takes 2 hours. I'd like to know if it is
normal to take all this time or no ? and also I'd like to know if I can
increase the performance

Thanks in advance for your answers!

Re: Jena Pellet Performance

[Diana Magdi <di...@gmail.com>]

Hi Dave,

Thanks for your replying

I loaded all the triple into memory and i did the reasoning by Pellet..
The time the reasoning took 01:33 min.

On 1 June 2015 at 20:11, Dave Reynolds <da...@gmail.com> wrote:

> On 01/06/15 18:00, rasha fawzy wrote:
>
>> Hi,
>>
>> i tried to load a file (20 588 triple) using jena tdb and to reason this
>> file by pellet but the time the reasoning takes is unacceptable. To return
>> the results of a simple query, it takes 2 hours. I'd like to know if it is
>> normal to take all this time or no ? and also I'd like to know if I can
>> increase the performance
>>
>> Thanks in advance for your answers!
>>
>
> As mentioned in another thread Pellet is not part of Apache Jena so you
> will need to ask the Pellet folks for support.
>
> However, as far as I know, Pellet will build all its inference structures
> in-memory so running over TDB will just result in slower performance it
> won't give you increased scalability of reasoning.
>
> I would start by loading all your triples into memory and seeing what
> Pellet performance is like.
>
> If you don't need complete DL inference than you may find the rule
> reasoners (e.g. OWLMicro running over an in-memory model) give you better
> performance at the cost of much weaker coverage.
>
> Dave
>
>
>
>
>

Re: Jena Pellet Performance

[Dave Reynolds <da...@gmail.com>]

On 01/06/15 18:00, rasha fawzy wrote:
> Hi,
>
> i tried to load a file (20 588 triple) using jena tdb and to reason this
> file by pellet but the time the reasoning takes is unacceptable. To return
> the results of a simple query, it takes 2 hours. I'd like to know if it is
> normal to take all this time or no ? and also I'd like to know if I can
> increase the performance
>
> Thanks in advance for your answers!

As mentioned in another thread Pellet is not part of Apache Jena so you 
will need to ask the Pellet folks for support.

However, as far as I know, Pellet will build all its inference 
structures in-memory so running over TDB will just result in slower 
performance it won't give you increased scalability of reasoning.

I would start by loading all your triples into memory and seeing what 
Pellet performance is like.

If you don't need complete DL inference than you may find the rule 
reasoners (e.g. OWLMicro running over an in-memory model) give you 
better performance at the cost of much weaker coverage.

Dave