9:53 a.m.
Hi
As some of you know, I'm using Drools as an engine for N3 logic (
http://www.w3.org/2000/10/swap/doc/Rules) .
N3 is a language equivalent to Semantic Web's RDF language, but much easier
to read and write by hand.
Moreover, N3 has, on top of RDF, an extension for rules, e.g. :
{ ?x parent ?y } => { ?y child ?x }.
Among many other vocabularies searchable on search site Swoogle (
http://swoogle.umbc.edu/), there is OWL, (OWL Web Ontology Language) that
allows to express in RDF, and thus in N3, rich models with classes,
properties, inheritance and constraints.
So in summary, N3 is a user-friendly language allowing to express facts,
class and properties, and rules.
So what does that bring for Drools users ?
One use case :
- open any number of RDF / OWL / N3 documents
- test and debug the rules using 3 rule engines (Drools, Euler, CWM)
- export all that project as a set of Drools packages, plus the facts in
XMLEncoder format ( by clicking on File / "Export as Drools" )
You can also use standard Semantic Web tools such as Protégé and Swoop to
edit and visualize RDF and OWL files.
Moreover, there are, as part of the Euler project, a library of N3 rules
that implement the logic of OWL and RDF Schema (transitive property,
inheritance, etc), and other goodies, see:
http://eulersharp.svn.sourceforge.net/viewvc/eulersharp/trunk/2003/03swap...
I translate N3 logic into Drools language, using just two classes
corresponding to Triple (statement) ( explanations here :
http://eulergui.svn.sourceforge.net/viewvc/eulergui/trunk/eulergui/html/d...).
Download from
https://sourceforge.net/projects/eulergui
Manual here:
http://eulergui.svn.sourceforge.net/viewvc/eulergui/trunk/eulergui/html/d...
Enjoy !
I've been working on this for several months, so it's beginning to
stabilize. But the rich set of built-in functions for other N3 engines is
still missing, with a few exceptions.
I'm using this framework currently to generate GUI's from domain models (
rules are here : http://deductions.svn.sourceforge.net/viewvc/deductions/n3/).
You will be glad to read that it runs around 2 seconds including
compilation, compared to the other engine (in Prolog) I use, which need
hundreds of seconds (actually a high complexity). A forward chaining engine
seems to make sense for such a constructive task.
I also wrote a How To about what I do currently with EulerGUI :
Deduction project - Java Swing application generator from OWL model and N3
logic rules
http://deductions.svn.sourceforge.net/viewvc/deductions/html/GUIgenerator...
Deduction project is work in progress.
--
Jean-Marc Vanel
http://jmvanel.free.fr/
6:57 a.m.
Hi Jean-Marc,
cc'ed to drools-resarch. Don't know where we want to lead this discussion...
Jean-Marc Vanel wrote:
As some of you know, I'm using Drools as an engine for N3 logic
(http://www.w3.org/2000/10/swap/doc/Rules) .
So in summary, N3 is a user-friendly language allowing to express
facts,
class and properties, and rules.
Nice to see some more work on integrating drools with the semantic web
side. As I have worked on a related topic recently:
- Do I understand you correctly, that you use Drools as rule engine for
rules + data where rules + data are defined in OWL?
- What standard are your rules based on?
- How capable is the OWL rules standard compared to Drools? What do you
think?
- What do you think about SPIN (http://spinrdf.org/)?
- Imho, the drools rule formulation are more user friendly and easier to
developed. Have you thought about running your data against native
drools rules (e. g. mvel dialect?)
cu, Michael
8:02 a.m.
New subject: EulerGUI released - IDE for semantic Web with Drools engine and Drools export
2009/2/24 Michael Zimmermann <list(a)incunabulum.de>
Hi Jean-Marc,
Jean-Marc Vanel wrote:
...
> So in summary, N3 is a user-friendly language allowing to express facts,
> class and properties, and rules.
Nice to see some more work on integrating drools with the semantic web
side. As I have worked on a related topic recently:
- Do I understand you correctly, that you use Drools as rule engine for
rules + data where rules + data are defined in OWL?
OWL is not a language for rules, it is a language for expressing complex
class definitions, e.g; "a vegetarian pizza is a pizza whose topping has no
meat" .
- What standard are your rules based on?
N3 logic : tutorial: http://www.w3.org/2000/10/swap/doc/Rules ;
W3C submission: http://www.w3.org/TeamSubmission/n3/#keywords
- How capable is the OWL rules standard compared to Drools? What do you
think?
First the language is more compact, second you don't need to define Pojo or
any classes beforehand. Let's take an example:
@prefix : <http://exampleS.com#> .
{ ?x :parent ?y } => { ?y :child ?x }.
Here is the Drools translation that you can run inside EulerGUI, and export
out of it :
rule "parents.n3 0"
when
Triple( $x : subject, predicate == "<http://exampleS.com#parent>", $y :
object )
then
insert( new Triple( $y, "<http://exampleS.com#child>", $x ) );
end
I would say that for pure logic rules (not too many Java calls in RHS) , and
moreover if you have some RDF data to process, and OWL class and property
definitions, it makes sense to use N3 rules.
EulerGUI offers 3 rules engines. The Drools translation is still lacking
many builtin functions that the Prolog EYE engine is offering (
http://www.w3.org/TR/xpath-functions/). Also the Prolog engine is able to
express logical or (disjunction) in the RHS .
On the other hand, for rules recursively building object graphs, the prolog
engine is much too slow.
In fact you can have the best of both worlds. Suppose you have a wealth of
Drools rules, plus also some RDF data and OWL ontologies. You can export
your RDF data and OWL ontologies as Java ojects ( just one class Triple is
used). Then you can write some Drools rules that express the relations
between the Triple objects from the RDF world and the other objects refered
to in the Drools rules.
In this scenario, it is notable that the logic of OWL and RDF Schema
(transitive property, inheritance, etc), is already expressed as N3 rules
(and hence as Drools rules ).
Hope this clarifies.
--
Jean-Marc Vanel
http://jmvanel.free.fr/
10:43 a.m.
Hi Jean-Marc,
Are you using the template functionality in Drools 5.0 for OWL? We
use Drools for expressing models/games down here and I have been
looking for an approach that would provide a more generic way of
defining data-models and rules that act on them.
Also, I wonder if you've looked at Miles Parker's MetaScape ABM
modeling plugin for Eclipse. It's a different area, but he has put
together a very nice visual way of working with both models (data) and
actions (rules).
I will download your project and look a bit deeper.
Many thanks for the posting and the links; wonderful discussion.
best wishes,
Andrew
---------------------------------
Andrew Waterman
San Cristóbal de las Casas, Chiapas, Mexico
+52 1 967 107 5902
+1 510 342 5693
On Feb 23, 2009, at 9:53 AM, Jean-Marc Vanel wrote:
Hi
As some of you know, I'm using Drools as an engine for N3 logic
(http://www.w3.org/2000/10/swap/doc/Rules
) .
N3 is a language equivalent to Semantic Web's RDF language, but much
easier to read and write by hand.
Moreover, N3 has, on top of RDF, an extension for rules, e.g. :
{ ?x parent ?y } => { ?y child ?x }.
Among many other vocabularies searchable on search site Swoogle (http://swoogle.umbc.edu/
), there is OWL, (OWL Web Ontology Language) that allows to express
in RDF, and thus in N3, rich models with classes, properties,
inheritance and constraints.
So in summary, N3 is a user-friendly language allowing to express
facts, class and properties, and rules.
So what does that bring for Drools users ?
One use case :
- open any number of RDF / OWL / N3 documents
- test and debug the rules using 3 rule engines (Drools, Euler, CWM)
- export all that project as a set of Drools packages, plus the
facts in XMLEncoder format ( by clicking on File / "Export as
Drools" )
You can also use standard Semantic Web tools such as Protégé and
Swoop to edit and visualize RDF and OWL files.
Moreover, there are, as part of the Euler project, a library of N3
rules that implement the logic of OWL and RDF Schema (transitive
property, inheritance, etc), and other goodies, see:
http://eulersharp.svn.sourceforge.net/viewvc/eulersharp/trunk/2003/03swap...
I translate N3 logic into Drools language, using just two classes
corresponding to Triple (statement) ( explanations here :
http://eulergui.svn.sourceforge.net/viewvc/eulergui/trunk/eulergui/html/d...
).
Download from
https://sourceforge.net/projects/eulergui
Manual here:
http://eulergui.svn.sourceforge.net/viewvc/eulergui/trunk/eulergui/html/d...
Enjoy !
I've been working on this for several months, so it's beginning to
stabilize. But the rich set of built-in functions for other N3
engines is still missing, with a few exceptions.
I'm using this framework currently to generate GUI's from domain
models ( rules are here : http://deductions.svn.sourceforge.net/viewvc/deductions/n3/
).
You will be glad to read that it runs around 2 seconds including
compilation, compared to the other engine (in Prolog) I use, which
need hundreds of seconds (actually a high complexity). A forward
chaining engine seems to make sense for such a constructive task.
I also wrote a How To about what I do currently with EulerGUI :
Deduction project - Java Swing application generator from OWL model
and N3 logic rules
http://deductions.svn.sourceforge.net/viewvc/deductions/html/GUIgenerator...
Deduction project is work in progress.
--
Jean-Marc Vanel
http://jmvanel.free.fr/
_______________________________________________
rules-users mailing list
rules-users(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-users
12:05 p.m.
New subject: EulerGUI released - IDE for semantic Web with Drools engine and Drools export
2009/2/24 Andrew Waterman <andrew.waterman(a)gmail.com>
Hi Jean-Marc,
Are you using the template functionality in Drools 5.0 for OWL? We use
Drools for expressing models/games down here and I have been looking for an
approach that would provide a more generic way of defining data-models and
rules that act on them.
No, you probably understood that I'm using from Drools "just" the inference
engine, and provide a different interface to it through the N3, RDF and OWL
inputs.
That being said, I had a look at template and I see common feature with what
I do, namely that :
"Decision tables also provide a degree of insulation from the underlying
object model"
which is very much the case with N3 rules. With N3 rules, you only really
care to define classes if you want to reason on classes themselves as data,
or if you want to reason on inheritance, or generic properties like
transitivity , etc .
Also, I wonder if you've looked at Miles Parker's MetaScape ABM modeling
plugin for Eclipse. It's a different area, but he has put
together a very
nice visual way of working with both models (data) and actions (rules).
I'll download and test it, but this is an approach based on eclipse and EMF,
like ATL, QVT and other stuff, which are interesting. And I've been working
2 years in a software company for an EMF like UML architecture, so I know
what I am talking about.
But I believe in the future of Logic and AI - based approches, as much
independent on infrastuctures as possible. And eclipse is definitely not a
light-weight infrastructure.
My goal with EulerGUI, and the larger project Deductions (
http://deductions.sourceforge.net/) is to be able to generate (to infer) as
much as possible of an application, on several platforms.
--
Jean-Marc Vanel
http://jmvanel.free.fr/
4:06 p.m.
Hi
Sorry for the inconvenience.
Since I uploaded it yesterday on Sourceforge, the EulerGUI release is *not
available* for download.
Sourceforge is unable to tell a time for solving the problem.
I'll let you know.
JMV
2009/2/23 Jean-Marc Vanel <jeanmarc.vanel(a)gmail.com>
...
Download from
https://sourceforge.net/projects/eulergui
Manual here:
http://eulergui.svn.sourceforge.net/viewvc/eulergui/trunk/eulergui/html/d...
--
Jean-Marc Vanel
http://jmvanel.free.fr/
--
Jean-Marc Vanel
http://jmvanel.free.fr/
7:10 p.m.
New subject: Guvnor Enumerations for integers?
I would like to have a enum like:
'Event.severity' : [''1=Information', '2=Warning',
'3=Critical']
Where Event.severity is a int.
However the guided rule builder puts my ints in ""
Like:
e : Event( severity > "1" )
are the enums supposed to work only with Strings?
thanks
--zoly
6:09 a.m.
2009/2/24 Jean-Marc Vanel <jeanmarc.vanel(a)gmail.com>
Sorry for the inconvenience.
Since I uploaded it yesterday on Sourceforge, the EulerGUI release is *not available* for
download.
Sourceforge is unable to tell a time for solving the problem.
I'll let you know.
With some help from Sourceforge, the download problem was solved ( I
verified the download ! ).
Download from
https://sourceforge.net/projects/eulergui
Manual here:
http://eulergui.svn.sourceforge.net/viewvc/eulergui/trunk/eulergui/html/d...
--
Jean-Marc Vanel
http://jmvanel.free.fr/
5752
days inactive
5755
days old
7 comments
4 participants
participants (4)
-
Andrew Waterman -
Jean-Marc Vanel -
Michael Zimmermann -
Zoltan Farkas