Hi

I somehow completely lose overview on "semantic" technologies...

Does anybody know or is willing to explain where the
difference/overlap... between "expert systems" like JESS/CLIPS and
ontologies (with reasoners) is. Because of the rules it appears that
expert systems are more powerful  but maybe there is something I can
express with ontologies that I can not within expert systems.

Another question is about RACER PRO. The query language NRQL looks like
a rule language in JESS. Is it RACER PRO's NRQL as powerful as the rules
in JESS? Maybe anybody knows both systems and could give a hint or a
link.

And last, does anybody know a way to import a OWL ontology into JESS.
Most solutions I know lose almost all restrictions. Is this an
implementation issue or is it because JESS (experts system in general)
are not to be used with "ontologies".

Thank you very much!
Gerhard
Martijn Tromm
Fri, 27 Oct 2006 16:17:47 -0700
Hi Gerhard,

I'm not very experienced with the technical of side of this, but I know 
something about rules and ontologies. Rules can only work on some datamodel. An 
ontology is nothing more than a formalized conceptualisation, so a extended 
datamodel. Extended with what? Wel with, constraints, typing, hierarchies, and 
rules. Expert systems can be seen as one of more possible implementations of 
logic within applications. And when that logic is highly irregular and prone to 
change than it is a good choice. OWL is one of more possible notations of 
ontologies. w3c is busy with standards for OWL rules (very similar to RuleML I 
believe). So rules and ontologies are a perfect couple. Expert systems shells 
are terrific middle ware for dealing with the rules that will be formulated 
upon ontologies. One of the big issues in this discussion is whether Owl 
applications should be Closed World applications. (OWL is not closed by 
default, but one can state a lot more rules under the CWA)

cheers,
Martijn
Samson Tu
Mon, 30 Oct 2006 14:49:41 -0800
The contrast is between description-logic (DL) reasoning (of which OWL is a variety) versus rule-based reasoning. Description-logic reasoning gives you runtime class-subsumption checking and classification of individuals based on necessary and sufficient definitions of classes. Rule-based reasoning is not designed to do these tasks. On the other hand, DL reasoning can't handle rules like (Father ?x ?y)(Brother ?x ?z) => (Uncle ?z ?y). 
OWL has been augmented with a rule-language called SWRL.

DL reasoning typically makes open-world assumption. Jess uses closed-world assumption. 
Making sense of DL restrictions (e.g., class expressions) typically requires the use of DL reasoner, which is difficult to simulate using something like Jess. 
Ontologies can be defined using DL language like OWL or frame-based language (e.g., Protege frame). Protege frame also makes closed-world assumption. Thus, JessTab works well integrating frame-based Protege ontologies with Jess. 
Samson
Gerhard Austaller
Tue, 31 Oct 2006 04:26:17 -0800
Hi

> -----Original Message-----
> From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]
On
> Behalf Of Samson Tu
> Sent: Monday, October 30, 2006 11:20 PM
> To: jess-users@sandia.gov
> Subject: Re: JESS: JESS vs RACER
> 
> 
> The contrast is between description-logic (DL) reasoning (of which OWL
> is a variety) versus rule-based reasoning. Description-logic reasoning
> gives you runtime class-subsumption checking and classification of
> individuals based on necessary and sufficient definitions of classes.
> Rule-based reasoning  is not designed to do these tasks. On the other
> hand, DL reasoning can't handle rules like (Father ?x ?y)(Brother ?x
?z)
> => (Uncle ?z ?y).
> 
> OWL has been augmented with a rule-language called SWRL.
> 
> DL reasoning typically makes open-world assumption. Jess uses
> closed-world assumption.

Thx for answer. OK. Now we are getting to the point. What I really want
to understand is where the difference is between rule "extensions" to
ontologies and rules from expert systems. What I'd like to is create new
individuals ("add a fact") in my ontology by using a rule language (like
SWRL). Of course this new individual my trigger other rules and so on
until a stable state. I'd like to program declarative as I am used to
with e.g. JESS.


> Making sense of DL restrictions (e.g., class expressions) typically
> requires  the use of DL reasoner, which is difficult to simulate using
> something like Jess.

But in principle there is no obstacle to merge the power of ontologies
(based on DL) and expert systems? 


> Ontologies can be defined using DL language like OWL or frame-based
> language (e.g., Protege frame). Protege frame also makes closed-world
> assumption. Thus, JessTab works well integrating frame-based Protege
> ontologies with Jess.

But the documentation of JessTab says that one OWL constraints and
expressions are not supported. So restrictions are not lost? 

Gerhard
Samson Tu
Tue, 31 Oct 2006 10:07:28 -0800
Gerhard Austaller wrote:
..
Thx for answer. OK. Now we are getting to the point. What I really want
to understand is where the difference is between rule "extensions" to
ontologies and rules from expert systems. What I'd like to is create new
individuals ("add a fact") in my ontology by using a rule language (like
SWRL). Of course this new individual my trigger other rules and so on
until a stable state. I'd like to program declarative as I am used to
with e.g. JESS.
I am not sure I can add more than what I said already. With rule extensions to DL languages, the rule system has to take into account the semantics of DL descriptions. To create new instances in SWRL, I think you need to have a "built-in" that does it. Creating new instances is not part of the language. 
Making sense of DL restrictions (e.g., class expressions) typically
requires  the use of DL reasoner, which is difficult to simulate using
something like Jess.
But in principle there is no obstacle to merge the power of ontologies
(based on DL) and expert systems? 
No, that's what SWRL and other semantic web rule languages are about.

Ontologies can be defined using DL language like OWL or frame-based
language (e.g., Protege frame). Protege frame also makes closed-world
assumption. Thus, JessTab works well integrating frame-based Protege
ontologies with Jess.
But the documentation of JessTab says that one OWL constraints and
expressions are not supported. So restrictions are not lost? 
Not sure what you mean. With OWL, JessTab doesn't 
Samson Tu
Tue, 31 Oct 2006 10:07:28 -0800
Gerhard Austaller wrote:
..
Thx for answer. OK. Now we are getting to the point. What I really want
to understand is where the difference is between rule "extensions" to
ontologies and rules from expert systems. What I'd like to is create new
individuals ("add a fact") in my ontology by using a rule language (like
SWRL). Of course this new individual my trigger other rules and so on
until a stable state. I'd like to program declarative as I am used to
with e.g. JESS.
I am not sure I can add more than what I said already. With rule extensions to DL languages, the rule system has to take into account the semantics of DL descriptions. To create new instances in SWRL, I think you need to have a "built-in" that does it. Creating new instances is not part of the language. 
Making sense of DL restrictions (e.g., class expressions) typically
requires  the use of DL reasoner, which is difficult to simulate using
something like Jess.
But in principle there is no obstacle to merge the power of ontologies
(based on DL) and expert systems? 
No, that's what SWRL and other semantic web rule languages are about.

Ontologies can be defined using DL language like OWL or frame-based
language (e.g., Protege frame). Protege frame also makes closed-world
assumption. Thus, JessTab works well integrating frame-based Protege
ontologies with Jess.
But the documentation of JessTab says that one OWL constraints and
expressions are not supported. So restrictions are not lost? 
Not sure what you mean. With OWL, JessTab doesn't take into account OWL constrains and expressions, but nothing is "losttake into account OWL constrains and expressions, but nothing is "lost
……
见http://www.mail-archive.com/jess-users@sandia.gov/msg08868.html