For discussions on Sowa's Conceptual Graphs, and its related topics e.g. Semantic Web, Knowledge Representation

Avner Levy | 1 Nov 2006 07:21

Re: Prolog Library for Handling CG Available?

>>> Any pointers to such libraries with source codes available

Mmmmh I am not sure. When I was in Europe (Marseilles) I heard of some language named Golog that seemed to have an implementation, but it was not really useable at that time. Maybe it is better now

Avner

Woon Kiong Tan <tanwoonk-/E1597aS9LQAvxtiuMwx3w@public.gmane.org> wrote:

Hi

Ulrik, Avner: Thanks for your pointers.

I'm looking more for a Prolog library that can run in Prolog systems like SWI-Prolog or Ciao. Any pointers to such libraries with source codes available? Thanks.

Regards,
Woon Kiong

Avner Levy <aglv02-/E1597aS9LQAvxtiuMwx3w@public.gmane.org> wrote:

Hi Ulrik,

That explains, I thought these were two different implementations with the same name. I know Dr. Kabbaj's work.

Thanks

Avner

Ulrik Petersen <ulrikp-GBFWfhuOQpx/SzgSGea1oA@public.gmane.org> wrote:

Hi Avner,

Avner Levy wrote:
> I realized you launched this effort with Jorgen Albretsen - am I right?
> Avner

Not quite. Prof. Dr. Adil Kabbaj, who also wrote the Amine-platform, is
the one who wrote Prolog+CG 2.0. The Amine Platform contains a much
improved Prolog+CG. I was the maintainer of the "legacy" version
(version 2.0) for a couple of years:

http://prologpluscg.sourceforge.net

Jørgen Albretsen will probably be taking over the maintainership of this
legacy version, though I will likely still be involved in the
maintenance process.

HTH

Ulrik Petersen

>
> */Ulrik Petersen /* wrote:
>
> Hi,
>
> Avner Levy wrote:
> > Hi
> >
> > >>> number one is better
> >
> > The second (sourceforge) link was broken when I tried to open it.
>
> It works again (for me).
>
>
> > */Ulrik Petersen /* wrote:
> >
> > Hi,
> >
> > Woon Kiong Tan wrote:
> > > Hi,
> > >
> > > I'm looking a Prolog library for handling CG. I read a report
> > > regarding CGPro API, but didn't manage to locate any source codes.
> > > Is there any available Prolog library that provides an API
> > similar to
> > > CGPro?
> >
> > Two options:
> >
> > 1) http://amine-platform.sourceforge.net/
> >
> > 2) http://prologpluscg.sourceforge.net/
> >
> > Number (1) is better.
> >
> > Ulrik Petersen
> >
>
> Regards,
>
> Ulrik Petersen
>
>
> To post a message, send mail to majordomo-ndqbGWLCXyQvU1jxblBdo2D2FQJk+8+b@public.gmane.org
> with the command 'subscribe cg' in the message body
> To unsubscribe, send mail to majordomo-ndqbGWLCXyQvU1jxblBdo2D2FQJk+8+b@public.gmane.org with
> the command 'unsubscribe cg' in the message body.
> See http://news.gmane.org/gmane.comp.ai.conceptualgraphs for the
> mailing list archive.
> See http://conceptualgraphs.org for the Conceptual Graph Home Page.
>
>
> ------------------------------------------------------------------------
> Access over 1 million songs - Yahoo! Music Unlimited Try it today.
>

To post a message, send mail to majordomo-ndqbGWLCXyQvU1jxblBdo2D2FQJk+8+b@public.gmane.org with the command 'subscribe cg' in the message body
To unsubscribe, send mail to majordomo-ndqbGWLCXyQvU1jxblBdo2D2FQJk+8+b@public.gmane.org with the command 'unsubscribe cg' in the message body.
See http://news.gmane.org/gmane.comp.ai.conceptualgraphs for the mailing list archive.
See http://conceptualgraphs.org for the Conceptual Graph Home Page.

@public.gmane.org>

We have the perfect Group for you. Check out the handy changes to Yahoo! Groups.

Everyone is raving about the all-new Yahoo! Mail.

Access over 1 million songs - Yahoo! Music Unlimited Try it today.

headers

Avner Levy | 1 Nov 2006 08:03

Re: What is practiccal use of relation type hierarchy?

Hello Aivars

Human extends animal as John wrote.

>>> So if Rover is a dog that lives in Canada, Rover inherits
from both Dog and Canadian. If Pete is a manager who lives
in Canada, he inherits from Human, Employee, and Canadian.

We eliminate those who live in Canada and don't have Canadian citizenship.

- 'Canadian' (or 'German', etc), is a citizenship, it is an attribute of a human (or an attribute of an animal, for regulations that may vary depending on the country).

- 'Manager' is a role.

These two (Canadian and manager) are attributes. On the other hand, a company can have many sorts of employees: some are clerks, some are managers, etc.

So a role such as manager can help differentiate between representatives of the same base class: Employee.

A manager can have Employee as supertype.

Now some company may have non-human employees (security companies may employ dogs).

Pete does inherit from Employee and Human. But I'd say that in John's example, Canadian is an attribute of human, not a class per se.

However, 'Human' can serve as supertype for a 'Citizen' class

I'd also like John's comment on this if he has time

Avner

Aivars Endzelis <aendzelis-gM/Ye1E23mwN+BqQ9rBEUg@public.gmane.org> wrote:

Heather and John,

I really appreciate your answers, thanks!

Aivars Endzelis.

To post a message, send mail to maj ordomo-ndqbGWLCXyQvU1jxblBdo2D2FQJk+8+b@public.gmane.org with the command 'subscribe cg' in the message body
To unsubscribe, send mail to majordomo-ndqbGWLCXyQvU1jxblBdo2D2FQJk+8+b@public.gmane.org with the command 'unsubscribe cg' in the message body.
See http://news.gmane.org/gmane.comp.ai.conceptualgraphs for the mailing list archive.
See http://conceptualgraphs.org for the Conceptual Graph Home Page.

Low, Low, Low Rates! Check out Yahoo! Messenger's cheap PC-to-Phone call rates.

headers

John F. Sowa | 1 Nov 2006 14:10

Re: What is practiccal use of relation type hierarchy?

Avner,

That little example was not intended to give a detailed
ontology of Canadian citizenship or business organization.
Its only purpose was to illustrate how one type may be
a subtype of many different supertypes.

If you want to read more about these and related issues,
I have gathered together several papers on the topic:

    http://www.jfsowa.com/ontology/guided.htm
    Guided Tour of Ontology

John Sowa

To post a message, send mail to majordomo@... with
the command 'subscribe cg' in the message body
To unsubscribe, send mail to majordomo@... with the
command 'unsubscribe cg' in the message body.
See http://news.gmane.org/gmane.comp.ai.conceptualgraphs for the mailing list archive.
See http://conceptualgraphs.org for the Conceptual Graph Home Page.

headers

Avner Levy | 1 Nov 2006 16:43

Re: What is practiccal use of relation type hierarchy?

John,

>>> That little example was not intended to give a detailed
ontology of Canadian citizenship or business organization.
Its only purpose was to illustrate how one type may be
a subtype of many different supertypes.

I gathered, it is obvious.

My point was to illustrate how a isA (Human) and a hasA property (citizenship) can let derive "Citizen".

Thanks for your pointer.

Avner

"John F. Sowa" <sowa-pmV1UhJe4grR7s880joybQ@public.gmane.org> wrote:

Avner,

That little example was not intended to give a detailed
ontology of Canadian citizenship or business organization.
Its only purpose was to illustrate how one type may be
a subtype of many different supertypes.

If you want to read more about these and related issues,
I have gathered together several papers on the topic:

http://www.jfsowa.com/ontology/guided.htm
Guided Tour of Ontology

John Sowa

To post a message, send mail to majordomo-ndqbGWLCXyQvU1jxblBdo2D2FQJk+8+b@public.gmane.org with the command 'subscribe cg' in the message body
To unsubscribe, send mail to majordomo-ndqbGWLCXyQvU1jxblBdo2D2FQJk+8+b@public.gmane.org with the command 'unsubscribe cg' in the message body.
See http://news.gmane.org/gmane.comp.ai.conceptualgraphs for the mailing list archive.
See http://conceptualgraphs.org for the Conceptual Graph Home Page.

Low, Low, Low Rates! Check out Yahoo! Messenger's cheap PC-to-Phone call rates.

headers

Christopher Menzel | 1 Nov 2006 16:43

Re: What is practiccal use of relation type hierarchy?

On Nov 1, 2006, at 1:03 AM, Avner Levy wrote:
> ...
> [John wrote:]
>> So if Rover is a dog that lives in Canada, Rover inherits
>> from both Dog and Canadian.  If Pete is a manager who lives
>> in Canada, he inherits from Human, Employee, and Canadian.
>
> We eliminate those who live in Canada and don't have Canadian  
> citizenship.

I think John was only using a simple example to make a point about  
the possibility of multiple supertypes.

> - 'Canadian' (or 'German', etc), is a citizenship, it is an  
> attribute of a human (or an attribute of an animal, for regulations  
> that may vary depending on the country).

In some representation systems, there is no difference between  
classes and attributes.  For instance, in a simple first-order  
language, both might be represented by 1-place predicates.  And even  
in systems that distinguish classes and attributes, the choice  
between representing a notion like "Canadian" as one or the other is  
entirely upon the purposes of the ontology designer.  There is no  
intrinsic fact of the matter.

>  - 'Manager' is a role.

Perhaps, if you have roles as part of your representation system.   
But even still, you might want 'Manager' to indicate a class if,  
e.g., you wanted a simple taxonomy of employees.

> These two (Canadian and manager) are attributes.

If you so choose.

> Now some company may have non-human employees (security companies  
> may employ dogs).

Depends on whether your representation of "employee" includes  
essential attributes that dogs might not have, e.g., a social  
security number.

> Pete does inherit from Employee and Human. But I'd say that in  
> John's example, Canadian is an attribute of human, not a class per se.

Right, YOU would say, i.e., for your purposes that's how you'd choose  
to represent "Canadian".  That's a perfectly reasonable choice.  But  
someone else in another context might choose differently.

Chris Menzel

To post a message, send mail to majordomo@... with
the command 'subscribe cg' in the message body
To unsubscribe, send mail to majordomo@... with the
command 'unsubscribe cg' in the message body.
See http://news.gmane.org/gmane.comp.ai.conceptualgraphs for the mailing list archive.
See http://conceptualgraphs.org for the Conceptual Graph Home Page.

headers

Avner Levy | 1 Nov 2006 22:39

Re: What is practiccal use of relation type hierarchy?

Contexts - right.

Can't the same be said about John's example?

Avner

Christopher Menzel <cmenzel-mRW4Vj+HBIk@public.gmane.org> wrote:

On Nov 1, 2006, at 1:03 AM, Avner Levy wrote:
> ...
> [John wrote:]
>> So if Rover is a dog that lives in Canada, Rover inherits
>> from both Dog and Canadian. If Pete is a manager who lives
>> in Canada, he inherits from Human, Employee, and Canadian.
>
> We eliminate those who live in Canada and don't have Canadian
> citizenship.

I think John was only using a simple example to make a point about
the possibility of multiple supertypes.

> - 'Canadian' (or 'German', etc), is a citizenship, it is an > attribute of a human (or an attribute of an animal, for regulations
> that may vary depending on the country).

In some representation systems, there is no difference between
classes and attributes. For instance, in a simple first-order
language, both might be represented by 1-place predicates. And even
in systems that distinguish classes and attributes, the choice
between representing a notion like "Canadian" as one or the other is
entirely upon the purposes of the ontology designer. There is no
intrinsic fact of the matter.

> - 'Manager' is a role.

Perhaps, if you have roles as part of your representation system.
But even still, you might want 'Manager' to indicate a class if,
e.g., you wanted a simple taxonomy of employees.

> These two (Canadian and manager) are attributes.

If you so choose.

> Now some company may have non-human employees (security companies
> may emplo y dogs).

Depends on whether your representation of "employee" includes
essential attributes that dogs might not have, e.g., a social
security number.

> Pete does inherit from Employee and Human. But I'd say that in
> John's example, Canadian is an attribute of human, not a class per se.

Right, YOU would say, i.e., for your purposes that's how you'd choose
to represent "Canadian". That's a perfectly reasonable choice. But
someone else in another context might choose differently.

Chris Menzel

To post a message, send mail to majordomo-ndqbGWLCXyQvU1jxblBdo2D2FQJk+8+b@public.gmane.org with the command 'subscribe cg' in the message body
To unsubscribe, send mail to majordomo-ndqbGWLCXyQvU1jxblBdo2D2FQJk+8+b@public.gmane.org with the command 'unsubscribe cg' in the message body.
See http://news.gmane.org/gmane.comp.ai.conceptualgraphs for the mailing list archive.
See http://conceptualgraphs.org for the Conceptual Graph Home Page.

QUOTE>

Everyone is raving about the all-new Yahoo! Mail.

headers

Avner Levy | 1 Nov 2006 23:03

Re: What is practiccal use of relation type hierarchy?

Let us limit this to my comment in a previous message:

>>> My point was to illustrate how a isA (Human) and a hasA property (citizenship) can let derive "Citizen".

Avner

Christopher Menzel <cmenzel-mRW4Vj+HBIk@public.gmane.org> wrote:

On Nov 1, 2006, at 1:03 AM, Avner Levy wrote:
> ...
> [John wrote:]
>> So if Rover is a dog that lives in Canada, Rover inherits
>> from both Dog and Canadian. If Pete is a manager who lives
>> in Canada, he inherits from Human, Employee, and Canadian.
>
> We eliminate those who live in Canada and don't have Cana dian
> citizenship.

I think John was only using a simple example to make a point about
the possibility of multiple supertypes.

> - 'Canadian' (or 'German', etc), is a citizenship, it is an
> attribute of a human (or an attribute of an animal, for regulations
> that may vary depending on the country).

In some representation systems, there is no difference between
classes and attributes. For instance, in a simple first-order
language, both might be represented by 1-place predicates. And even
in systems that distinguish classes and attributes, the choice
between representing a notion like "Canadian" as one or the other is
entirely upon the purposes of the ontology designer. There is no
intrinsic fact of the matter.

> - 'Manager' is a role.

Perhaps, if you have roles as part of your representation system.
But even still, you might want 'Manager' to indicate a class if,
e.g., you wa nted a simple taxonomy of employees.

> These two (Canadian and manager) are attributes.

If you so choose.

> Now some company may have non-human employees (security companies
> may employ dogs).

Depends on whether your representation of "employee" includes
essential attributes that dogs might not have, e.g., a social
security number.

> Pete does inherit from Employee and Human. But I'd say that in
> John's example, Canadian is an attribute of human, not a class per se.

Right, YOU would say, i.e., for your purposes that's how you'd choose
to represent "Canadian". That's a perfectly reasonable choice. But
someone else in another context might choose differently.

Chris Menzel

To post a message, send mail to majordomo-ndqbGWLCXyQvU1jxblBdo2D2FQJk+8+b@public.gmane.org with the command 'subscribe cg' in the message body
To unsubscribe, send mail to majordomo-ndqbGWLCXyQvU1jxblBdo2D2FQJk+8+b@public.gmane.org with the command 'u nsubscribe cg' in the message body.
See http://news.gmane.org/gmane.comp.ai.conceptualgraphs for the mailing list archive.
See http://conceptualgraphs.org for the Conceptual Graph Home Page.

Check out the New Yahoo! Mail - Fire up a more powerful email and get things done faster.

headers

Christopher Menzel | 2 Nov 2006 05:42

Re: What is practiccal use of relation type hierarchy?

> Contexts - right.
>
> Can't the same be said about John's example?

To be sure!  

To post a message, send mail to majordomo@... with
the command 'subscribe cg' in the message body
To unsubscribe, send mail to majordomo@... with the
command 'unsubscribe cg' in the message body.
See http://news.gmane.org/gmane.comp.ai.conceptualgraphs for the mailing list archive.
See http://conceptualgraphs.org for the Conceptual Graph Home Page.

headers

John F. Sowa | 2 Nov 2006 06:05

Re: What is the practical use of relation type hierarchy?

Avner and Chris,

AL>> Contexts - right.
 >>
 >> Can't the same be said about John's example?

CM>> To be sure!

Yes, everything about natural language is context
dependent -- phonology, syntax, semantic, & pragmatics.

AL> My point was to illustrate how a isA (Human) and a hasA
> property (citizenship) can let derive "Citizen".

You could state the following axiom:

    For every human x, if x has a property of citizenship,
    then x is a citizen.

In predicate calculus, that could be written

    (Ax:Human)(hasProperty(x,citizenship) -> Citizen(x)).

But then you would have to write more axioms about the
implications of Citizen(x).   You would also have to
address the question about how to distinguish the
country of citizenship, and that would mean you have
a property-function rather than just a property.

I'll illustrate some of the issues by stating axioms for
Manager(x), which is a somewhat simpler relation than
citizenship.

As Chris also mentioned, the word _manager_ could be
represented formally in many different ways.  One way is
by a dyadic predicate managerOf(x,y), which says that some
person x is the manager of some person y.

However, it is also common to represent a noun by a monadic
predicate, such as Manager(x), which is true of anybody who
is a manager.  The associated relations would be defined
by axioms such as the following:

    1. Every person who is employed by a company c
       is an employee of c.

    2. Every manager x is employed by some company c,
       and x is a manager of c.

    3. For every manager x of c, there exists a nonempty
       set S of people who are employed by c.

    4. Every employee y in S reports to the manager x.

    5. There exists exactly one CEO of c, who is a manager
       of c and who does not report to any manager of c.

    6. Every employee x of c except for the CEO of c
       reports to exactly one manager y of c.

    7. No manager x of c reports directly or indirectly to x.

For readability, I stated these axioms in a stylized version
of English, which could be translated to whatever version
of logic you prefer.

John

To post a message, send mail to majordomo@... with
the command 'subscribe cg' in the message body
To unsubscribe, send mail to majordomo@... with the
command 'unsubscribe cg' in the message body.
See http://news.gmane.org/gmane.comp.ai.conceptualgraphs for the mailing list archive.
See http://conceptualgraphs.org for the Conceptual Graph Home Page.

headers

John F. Sowa | 2 Nov 2006 15:22

Guided Tour of Ontology

I recently received a request for a copy of the conceptual
catalog at the end of my 1984 book, which I updated and
extended for the 2000 book.  Following is my response,
which may be of interest to other readers of this list:

I recommend the Guided Tour of Ontology, which contains
several papers plus the material in Appendix B of the
KR book (which is a superset of the old Appendix B).

    http://www.jfsowa.com/ontology/guided.htm

Go to Section 5, KR Ontology, and you will find six
subsections.  Each one covers part of the ontology.

John Sowa

To post a message, send mail to majordomo@... with
the command 'subscribe cg' in the message body
To unsubscribe, send mail to majordomo@... with the
command 'unsubscribe cg' in the message body.
See http://news.gmane.org/gmane.comp.ai.conceptualgraphs for the mailing list archive.
See http://conceptualgraphs.org for the Conceptual Graph Home Page.

Mon	Tue	Wed	Thu	Fri	Sat	Sun

		1	2	3	4	5
6	7	8	9	10	11	12
13	14	15	16	17	18	19
20	21	22	23	24	25	26
27	28	29	30

1	May 2013
1	March 2013
15	February 2013
13	January 2013
9	December 2012
5	November 2012
5	October 2012
3	September 2012
5	August 2012
22	July 2012
10	June 2012
8	May 2012
10	April 2012
24	March 2012
8	February 2012
15	January 2012
26	December 2011
27	November 2011
10	October 2011
4	September 2011
2	August 2011
5	July 2011
1	June 2011
14	May 2011
9	April 2011
13	March 2011
21	February 2011
13	January 2011
7	December 2010
10	November 2010
8	September 2010
20	August 2010
1	July 2010
1	April 2010
1	March 2010
38	January 2010
91	December 2009
34	November 2009
25	October 2009
16	September 2009
19	August 2009
5	July 2009
26	June 2009
29	May 2009
7	April 2009
47	March 2009
12	February 2009
11	January 2009
73	December 2008
7	November 2008
16	October 2008
32	September 2008
76	August 2008
17	July 2008
18	June 2008
22	May 2008
35	April 2008
22	March 2008
45	February 2008
20	January 2008
28	December 2007
35	November 2007
3	October 2007
1	August 2007
68	July 2007
63	June 2007
57	May 2007
40	April 2007
106	March 2007
18	February 2007
14	January 2007
24	December 2006
29	November 2006
32	October 2006
14	September 2006
54	August 2006
29	July 2006
92	June 2006
84	May 2006
68	April 2006
19	March 2006
131	February 2006
258	January 2006
59	December 2005
139	November 2005
77	October 2005
197	September 2005
197	August 2005
57	July 2005
50	June 2005
24	May 2005
53	April 2005
83	March 2005
36	February 2005
7	January 2005