Hello Rodrigo,
the explanations in Protege are computed by using the API from Matthew
Horridge, which is pretty easy to use and implements his brilliant PhD
thesis.
The project is located here:
https://github.com/matthewhorridge/owlexplanation
Kind regards,
Lorenz
Hi Nicolas,
Don't forget to read the pellet faq about extracting inferences using jena:
https://github.com/Complexible/pellet/wiki/FAQ#how-can-i-extract-all-inferences
About the explanations, I'm not 100% sure, but I believe the methods used
to get them are from Pellet ( pelletReasoner.getKB().getExplanation() ),
so, they should work with both APIs. When you download pellet zip package,
there are some examples... at least one is about explanations...
Most of things you do with Jena you can do with OWLAPI and also the other
way around... both are quite simple to understand and a few lines of code
can do many things =)
I was able to implement my custom functions to use with SWRL, if you need
any help, I'll be glad to help!
Rodrigo C. Antonialli
======================================
Rio Claro - SP - Brasil
LinkedIn: http://www.linkedin.com/in/rcantonialli
Contato: [email protected]
Skype: rc_antonialli
On Tue, Dec 8, 2015 at 10:28 PM, Nicolas Paris <[email protected]> wrote:
Hi there,
I am working on these things too these days. I try to do all without
owlapi, and I hope I will be able to. Thank you all for these
question/clear answers.
About my tests, as far as I can see, with jena & pellet I am able to :
- transform the data from TDB and pass the pellet rules + knowlege
base on it, in 4 lines code.
- add individuals, data/objects properties and all OWL stuff thanks to
Jena OR pellet methods
- create SPARQL with jenaARQ OR SPARQLDL with Pellet
What I will try to do in the next days will be :
- get what you called the "InferredIndividualAxiomGenerator" within
jena. -> I have some starting points
(
http://stackoverflow.com/questions/25055263/how-to-get-only-inferred-data-from-jena-ontology-model
)
- get the explanations of inferred axioms from pellet (the same as
Protégé shows when you hit the "?") -> any idea ?.
- create my own pellet built-in (begining with implementing some dates
related built-in). Maybe make a pull request on github then -> I have
already read an interesting post of Rodrigo about that on the Pellet
forum.
2015-12-09 0:23 GMT+01:00 Rodrigo Antonialli <[email protected]>:
Short answer: yes, that's it...
Long answer:
I've created the ontologies using Protege, and uploaded them to Fuseki.
When I need the ontologies in my application, Jena is used to load the
ontologies from Fuseki. In most cases, I do things with Jena, specially
SPARQL.
But once the ontologies are stored into OntModel, they can be written to
some OutputStream and read from OWLAPI. Using OWLAPI at certain
application
features, I'm creating some individuals and running Pellet inference,
including SWRL rules (which, in my case, are defined in a separated
file/graph, which is better for maintenance).
There are 'tools' from OWLAPI to extract the inferred axioms only (like
InferredIndividualAxiomGenerator). In my case, I'm joining the inferred
axioms with the individuals and "exporting" back to Jena... from Jena
it's
just to upload them to Fuseki again...
That may looks like an ugly and unnecessary overhead from some point of
views.... but with that "architecture", I can benefit from both: Jena +
Fuseki (storage and SPARQL) and OWLAPI + Pellet (Inferences, SWRL)...
actually, it's a good way (or at least the one I've found) to use the
power
of both APIs. IMHO, there are some things easier to be done with Jena and
others easier with OWLAPI (mostly because they are focused on different
aspects... Jena is more "RDF-centric" and OWLAPI, "OWL-centric")
Rodrigo C. Antonialli
======================================
Rio Claro - SP - Brasil
LinkedIn: http://www.linkedin.com/in/rcantonialli
Contato: [email protected]
Skype: rc_antonialli
On Tue, Dec 8, 2015 at 7:17 PM, Chris Snyder <[email protected]>
wrote:
That is very helpful feedback.
As a clarification I attempted to use the Pellet code from Maven
central
and that is where I ran into the package name change issue. I might try
getting in touch with Ignazio directly to see if I can help make the
updates to get Pellet working with the latest Jena.
If you don’t mind my asking, how are you applying the SWRL rules in
practice? E.G. are you applying the SWRL rules using the OWLAPI +
Pellet
and then exporting the additional reasoned triples to your Fuseki
instance
so they can be queried with SPARQL?
Thanks,
Chris
On Dec 8, 2015, at 3:55 PM, Rodrigo Antonialli <
[email protected]>
wrote:
Hi Chris,
Maybe I'm not an expert to answer you properly, but I've being
working
with
this kind of scenario (Ontologies, SWRL, jena, owlapi) lately and
here is
what I can share:
1) Yes. As far as I know, Jena doesn't have native support for SWRL.
2) The most recent versions of pellet I've seen is the one from
Ignazio
Palmisano, which you can find at Maven Central
<http://search.maven.org/#search%7Cga%7C1%7Cignazio1977>. I know it
works
with most recent versions of OWL-API, can't tell what Jena version it
supports now... (I don't know if he is in this mailling list, but
you can
find him in owlapi mailling list =P)
3) Can't tell for sure, once I've never used Jena Rules, but I
believe
the
differences should be around details and most things would be made
with
both. Actually, Jena Rules and SWRL are "competing" (emphasis on the
quotes) with SPARQL SPIN from TopBraid, so it's difficult to find
clear
documented comparisons. There are a lot of considerations around the
subject...
4) The problem with other reasoners is, usually, not Jena
compatibility,
but SWRL support... As far as I could look around, pellet is the only
free/open-source reasoner with a good support for SWRL. I've spent
some
time looking around, because the pellet performance with SWRL is a
little
problematic when the ontology is complex, there are a great number of
rules... Although one thing I've found to be a nice feature is the
ability
to create custom functions for SWRL using Pellet. It was good for me
to
work with geospatial data (and now I'm trying to speed things up)...
5) "What should I use" questions are like "What's the best", and the
best
answer is: it depends... of a lot of things... I've being working
with
both
APIs and by now, I chose to use Jena to work with Fusek and SPARQL
and
OWLAPI to manipulate the ontology directly (build and manage axioms).
Simple SPARQL queries you can change for some OWLAPI code, but the
endpoint/tdb support in jena are an advantage. It's quite easy to
work
with
both in the same application (both API read and write from various
formats)
Rodrigo C. Antonialli
======================================
Rio Claro - SP - Brasil
LinkedIn: http://www.linkedin.com/in/rcantonialli
Contato: [email protected]
Skype: rc_antonialli
On Tue, Dec 8, 2015 at 6:18 PM, Chris Snyder <[email protected]>
wrote:
I have a few reasoner questions.
1) I’m using the latest version of Jena 3.0.0 is it true that the
only
way
to use SWRL rules is with an external reasoner? My concern is how I
would
use Protege which can handle adding SWRL rules and then processing
the
ontology with Jena.
2) I tried to integrate the Pellet reasoner but to get it to work
with
Jena 3.0.0 I believe I would need to fork the Pellet code and
update all
the Jena package paths in it to use the new “org.apache.jena.”
location.
Can anyone comment on that?
3) If I am using the built in Jena reasoner is there anything major
that I
can not do with the Jena rule language that I could do with SWRL?
4) Is there an reasoner other than Pellet that someone has used with
Jena
3.0.0 successfully that will process SWRL rules?
5) A reasoner is obviously an important part of using an ontology.
Is
Jena
the appropriate way to go or should I be looking into the OWLAPI?
Thanks in advance,
Chris
--
Lorenz Bühmann
AKSW group, University of Leipzig
Group: http://aksw.org - semantic web research center
