|
List Info
Thread: ActiveRDF and OWL support
|
|
| ActiveRDF and OWL support |
 
Ireland |
2007-05-22 06:17:25 |
Hi,
I am having a little bit of trouble with my eMail setup, but
basically this is meant as a reply for Damaris:
Currently there is only one easy option for getting OWL
support in
ActiveRDF:
use SPARQL to access an endpoint which does its own
reasoning.
The endpoint could be running:
* Jena, because Jena currently (according to their homepage)
supports
SPARQL and OWL.
* Pellet, according to homepage is a stand alone OWL
resoner, which
supports SPARQL (but I cant see a description of how to get
the data
into Pellet)
* You might want to look at ways to get OWL reasoned data
out of
protege, but the information I could find was all very
fuzzy
What you cant use right now, unless you develop it
yourself:
* Sesame 2, because it has no OWL support
* Sesame 1, because it has no SPARQL support. You could try
to write
a SeRQL adapter, then connect to a Sesame 1 server which
runs OWLIM
* OpenLink virtuoso: no OWL support
I hope these pointers help you.
cheers, Benjamin.
_______________________________________________
ActiveRDF mailing list
ActiveRDF lists.deri.org
http
://lists.deri.org/mailman/listinfo/activerdf
|
|
| Re: ActiveRDF and OWL support |
|
2007-05-22 13:46:35 |
On 05/22/07/05/07 12:17 +0100, Benjamin Heitmann wrote:
>What you cant use right now, unless you develop it
yourself:
>* Sesame 2, because it has no OWL support
>* Sesame 1, because it has no SPARQL support. You could
try to write
OWLIM integrates into Sesame and gives you OWL reasoning
(between OWL Lite
and OWL DL).
-eyal
_______________________________________________
ActiveRDF mailing list
ActiveRDFlists.deri.org
http
://lists.deri.org/mailman/listinfo/activerdf
|
|
| Re: ActiveRDF and OWL support |
Ireland |
2007-05-23 08:04:20 |
On May 22, 2007, at 7:46 PM, Eyal Oren wrote:
> On 05/22/07/05/07 12:17 +0100, Benjamin Heitmann
wrote:
>> What you cant use right now, unless you develop it
yourself:
>> * Sesame 2, because it has no OWL support
>> * Sesame 1, because it has no SPARQL support. You
could try to write
> OWLIM integrates into Sesame and gives you OWL
reasoning (between
> OWL Lite and OWL DL).
yes, but AFAIK OWLIM only works with Sesame 1, which does
not support
SPARQL, so the SPARQL route to access OWLIM is closed.
_______________________________________________
ActiveRDF mailing list
ActiveRDFlists.deri.org
http
://lists.deri.org/mailman/listinfo/activerdf
|
|
| Re: ActiveRDF and OWL support |
|
2007-05-23 10:20:41 |
|
Our group is learning Ruby + ActiveRDF at the same time, targeting to develop a site with faceted browsing + Ajax.
Currently we have "small bits" working, as we're trying different things out. After playing around with Sesame (with Perl scripts) we ended up using Jena as the RDF/OWL backend. We're using the Joseki server and the SPARQL adapter.
The biggest problem we're having at the moment is connecting a reasoner to Joseki. We still have no reasoning, although using Pellet or Jena's internal reasoner should be possible with Joseki. Should I mention, without the DIG protocol. One member of our team is working on it.
On activerdf_rules: it is slow.  For our little ontology and a handful of instances it took 10 minutes 44 seconds to process the RDF graph.
Ok, now to the interesting point. We hired a Java-compliant student over the summer to help us out. We would love to bypass the SPARQL server and use Jena's magic straight from Rails / CRuby. Ah, well. How are we actually going to use Java from CRuby? Any helpful hints are appreciated.. He's prepared to write an ActiveRDF adapter for Jena in JRuby. Benjamin, if you're reading, I think he'll need your help in a few weeks!
I believe this solution would - so to speak - rock!! I'd love to hear your opinions. CRuby and JRuby would run as a separate processes, and talked to each other through an interface. Best if they could pass a variable in native Ruby. JRuby would use the Jena adapter for queries, and could utilize the Jena model API and fancy features that are not available due to SPARQL limitations. Not to mention the speed boost. Jena could be linked to an external reasoner such as Pellet or RACER.
I hope this sparks some interest. Oh, and thanks for the fabulous gem called ActiveRDF!
Mikael
2007/5/22, Benjamin Heitmann < benjamin.heitmannderi.org">
benjamin.heitmannderi.org>:Hi,
I am having a little bit of trouble with my eMail setup, but
basically this is meant as a reply for Damaris:
Currently there is only one easy option for getting OWL support in
ActiveRDF:
use SPARQL to access an endpoint which does its own reasoning.
The endpoint could be running:
* Jena, because Jena currently (according to their homepage) supports
SPARQL and OWL.
* Pellet, according to homepage is a stand alone OWL resoner, which
supports SPARQL (but I cant see a description of how to get the data
into Pellet)
* You might want to look at ways to get OWL reasoned data out of
protege, but the information I could find was all very fuzzy
What you cant use right now, unless you develop it yourself:
* Sesame 2, because it has no OWL support
* Sesame 1, because it has no SPARQL support. You could try to write
a SeRQL adapter, then connect to a Sesame 1 server which runs OWLIM
* OpenLink virtuoso: no OWL support
I hope these pointers help you.
cheers, Benjamin.
_______________________________________________
ActiveRDF mailing list
ActiveRDFlists.deri.org">ActiveRDFlists.deri.org
http://lists.deri.org/mailman/listinfo/activerdf
|
| Accessing Sesame 1 java interface in
Ruby |
 
Brazil |
2007-05-23 10:33:18 |
Hi all,
as a reminder, we have used the Ruby Java Bridge (RJB) to
access Sesame
1's java interface directly, in our HyperDE environment (a
predecessor
in some sense of ActiveRDF): http://www.t
ecweb.inf.puc-rio.br/hyperde
Cheers,
D.
> Ok, now to the interesting point. We hired a
Java-compliant student
> over the summer to help us out. We would love to bypass
the SPARQL
> server and use Jena's magic straight from Rails /
CRuby. Ah, well. How
> are we actually going to use Java from CRuby? Any
helpful hints are
> appreciated.. He's prepared to write an ActiveRDF
adapter for Jena in
> JRuby. Benjamin, if you're reading, I think he'll need
your help in a
> few weeks!
>
> I believe this solution would - so to speak - rock!!
I'd love to hear
> your opinions. CRuby and JRuby would run as a separate
processes, and
> talked to each other through an interface. Best if they
could pass a
> variable in native Ruby. JRuby would use the Jena
adapter for queries,
> and could utilize the Jena model API and fancy features
that are not
> available due to SPARQL limitations. Not to mention the
speed boost.
> Jena could be linked to an external reasoner such as
Pellet or RACER.
_______________________________________________
ActiveRDF mailing list
ActiveRDFlists.deri.org
http
://lists.deri.org/mailman/listinfo/activerdf
|
|
| Re: ActiveRDF and OWL support |
|
2007-05-23 10:51:22 |
|
Mikael,
I'm the person who wrote activerdf-rules. I haven't done much with it since I wrote it initially. I'm not sure if there is a place for it or not. It was kind of designed to be used in situations with smallish databases. I think the better solution is to have a reasoning system in your triple store, especially for any serious applications.
Currently, activerdf-rules is horribly inefficient. It will process every rule, and every set of triples that satisfy every rule anytime a single triple is added to the store, even if the result of the rule already exists in the database. It could probably be modified to only search for rules where the triples that match the rule must at least contain the triple being added.
Also it uses forward chaining and it might perform better using backward chaining.
Finally, it could possibly be rewritten as a C extension.
It was kind of a one-off hack, and could be improved significantly with some simple modifications, but in the end it would take a lot of work to get it to perform decently. And like I said, I figured that a better solution is to have a reasoning triple store. I didn't think anyone had even tried activerdf-rules.
If there are people interested in using activerdf-rules, then let me know because I could be persuaded to do some more work on it. I figured there was no interest so it was dead. Additionally, if anyone is interested in helping to improve activerdf-rules so that it works well enough to use in simple situations, then contact me.
Paul
On 5/23/07, Mikael Lammentausta < mikael.lammentaustagmail.com">mikael.lammentaustagmail.com> wrote:
Our group is learning Ruby + ActiveRDF at the same time, targeting to develop a site with faceted browsing + Ajax.
Currently we have "small bits" working, as we're trying different things out. After playing around with Sesame (with Perl scripts) we ended up using Jena as the RDF/OWL backend. We're using the Joseki server and the SPARQL adapter.
The biggest problem we're having at the moment is connecting a reasoner to Joseki. We still have no reasoning, although using Pellet or Jena's internal reasoner should be possible with Joseki. Should I mention, without the DIG protocol. One member of our team is working on it.
On activerdf_rules: it is slow. For our little ontology and a handful of instances it took 10 minutes 44 seconds to process the RDF graph.
Ok, now to the interesting point. We hired a Java-compliant student over the summer to help us out. We would love to bypass the SPARQL server and use Jena's magic straight from Rails / CRuby. Ah, well. How are we actually going to use Java from CRuby? Any helpful hints are appreciated.. He's prepared to write an ActiveRDF adapter for Jena in JRuby. Benjamin, if you're reading, I think he'll need your help in a few weeks!
I believe this solution would - so to speak - rock!! I'd love to hear your opinions. CRuby and JRuby would run as a separate processes, and talked to each other through an interface. Best if they could pass a variable in native Ruby. JRuby would use the Jena adapter for queries, and could utilize the Jena model API and fancy features that are not available due to SPARQL limitations. Not to mention the speed boost. Jena could be linked to an external reasoner such as Pellet or RACER.
I hope this sparks some interest. Oh, and thanks for the fabulous gem called ActiveRDF!
Mikael
2007/5/22, Benjamin Heitmann < benjamin.heitmannderi.org" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">
benjamin.heitmannderi.org>:Hi,
I am having a little bit of trouble with my eMail setup, but
basically this is meant as a reply for Damaris:
Currently there is only one easy option for getting OWL support in
ActiveRDF:
use SPARQL to access an endpoint which does its own reasoning.
The endpoint could be running:
* Jena, because Jena currently (according to their homepage) supports
SPARQL and OWL.
* Pellet, according to homepage is a stand alone OWL resoner, which
supports SPARQL (but I cant see a description of how to get the data
into Pellet)
* You might want to look at ways to get OWL reasoned data out of
protege, but the information I could find was all very fuzzy
What you cant use right now, unless you develop it yourself:
* Sesame 2, because it has no OWL support
* Sesame 1, because it has no SPARQL support. You could try to write
a SeRQL adapter, then connect to a Sesame 1 server which runs OWLIM
* OpenLink virtuoso: no OWL support
I hope these pointers help you.
cheers, Benjamin.
_______________________________________________
ActiveRDF mailing list
ActiveRDFlists.deri.org" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">
ActiveRDFlists.deri.org
http://lists.deri.org/mailman/listinfo/activerdf
_______________________________________________
ActiveRDF mailing list
lists.deri.org">ActiveRDFlists.deri.org
http://lists.deri.org/mailman/listinfo/activerdf
|
| Re: ActiveRDF and OWL support |
|
2007-05-24 03:55:28 |
|
Hi Paul,
currently we need just simple subClassOf-relations, eventually more sophisticated property reasoning and that for there are many options. Got opinions on Pellet/FACT++/RACER/etc?
Honestly, I couldn';t figure it out how I could write my own rule base for activerdf_rules. It would be good to have the option to use the barest of all, bottom-to-top subClassOf relation rules.
Mikael
2007/5/23, Paul Stadig < paulstadig.name">paulstadig.name>:
Mikael,
I'm the person who wrote activerdf-rules. I haven't done much with it since I wrote it initially. I'm not sure if there is a place for it or not. It was kind of designed to be used in situations with smallish databases. I think the better solution is to have a reasoning system in your triple store, especially for any serious applications.
Currently, activerdf-rules is horribly inefficient. It will process every rule, and every set of triples that satisfy every rule anytime a single triple is added to the store, even if the result of the rule already exists in the database. It could probably be modified to only search for rules where the triples that match the rule must at least contain the triple being added.
Also it uses forward chaining and it might perform better using backward chaining.
Finally, it could possibly be rewritten as a C extension.
It was kind of a one-off hack, and could be improved significantly with some simple modifications, but in the end it would take a lot of work to get it to perform decently. And like I said, I figured that a better solution is to have a reasoning triple store. I didn't think anyone had even tried activerdf-rules.
If there are people interested in using activerdf-rules, then let me know because I could be persuaded to do some more work on it. I figured there was no interest so it was dead. Additionally, if anyone is interested in helping to improve activerdf-rules so that it works well enough to use in simple situations, then contact me.
Paul
On 5/23/07, Mikael Lammentausta < mikael.lammentaustagmail.com" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">
mikael.lammentaustagmail.com> wrote:
Our group is learning Ruby + ActiveRDF at the same time, targeting to develop a site with faceted browsing + Ajax.
Currently we have "small bits" working, as we're trying different things out. After playing around with Sesame (with Perl scripts) we ended up using Jena as the RDF/OWL backend. We're using the Joseki server and the SPARQL adapter.
The biggest problem we're having at the moment is connecting a reasoner to Joseki. We still have no reasoning, although using Pellet or Jena's internal reasoner should be possible with Joseki. Should I mention, without the DIG protocol. One member of our team is working on it.
On activerdf_rules: it is slow. For our little ontology and a handful of instances it took 10 minutes 44 seconds to process the RDF graph.
Ok, now to the interesting point. We hired a Java-compliant student over the summer to help us out. We would love to bypass the SPARQL server and use Jena's magic straight from Rails / CRuby. Ah, well. How are we actually going to use Java from CRuby? Any helpful hints are appreciated.. He's prepared to write an ActiveRDF adapter for Jena in JRuby. Benjamin, if you're reading, I think he'll need your help in a few weeks!
I believe this solution would - so to speak - rock!! I'd love to hear your opinions. CRuby and JRuby would run as a separate processes, and talked to each other through an interface. Best if they could pass a variable in native Ruby. JRuby would use the Jena adapter for queries, and could utilize the Jena model API and fancy features that are not available due to SPARQL limitations. Not to mention the speed boost. Jena could be linked to an external reasoner such as Pellet or RACER.
I hope this sparks some interest. Oh, and thanks for the fabulous gem called ActiveRDF!
Mikael
2007/5/22, Benjamin Heitmann < benjamin.heitmannderi.org" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">
benjamin.heitmannderi.org>:Hi,
I am having a little bit of trouble with my eMail setup, but
basically this is meant as a reply for Damaris:
Currently there is only one easy option for getting OWL support in
ActiveRDF:
use SPARQL to access an endpoint which does its own reasoning.
The endpoint could be running:
* Jena, because Jena currently (according to their homepage) supports
SPARQL and OWL.
* Pellet, according to homepage is a stand alone OWL resoner, which
supports SPARQL (but I cant see a description of how to get the data
into Pellet)
* You might want to look at ways to get OWL reasoned data out of
protege, but the information I could find was all very fuzzy
What you cant use right now, unless you develop it yourself:
* Sesame 2, because it has no OWL support
* Sesame 1, because it has no SPARQL support. You could try to write
a SeRQL adapter, then connect to a Sesame 1 server which runs OWLIM
* OpenLink virtuoso: no OWL support
I hope these pointers help you.
cheers, Benjamin.
_______________________________________________
ActiveRDF mailing list
ActiveRDFlists.deri.org" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">
ActiveRDFlists.deri.org
http://lists.deri.org/mailman/listinfo/activerdf
_______________________________________________
ActiveRDF mailing list
ActiveRDFlists.deri.org" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">ActiveRDFlists.deri.org
http://lists.deri.org/mailman/listinfo/activerdf
_______________________________________________
ActiveRDF mailing list
lists.deri.org">ActiveRDFlists.deri.org
http://lists.deri.org/mailman/listinfo/activerdf
|
[1-7]
|
|