[Modeling] Re: [Modeling] Agent-oriented: why is it necessary?
Danny Weyns
Danny.Weyns@cs.kuleuven.ac.be
Thu, 20 Nov 2003 15:50:42 +0100
--------------968E56FCC7A35F152AC941E9
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: 8bit
Hello,
Concepts and implementation should not be confused but agents systems
have to be build! So far agent research has mainly focussed on
conceptualization. Conceptualization intends to clarify what the
building blocks of multi-agent systems are and how they relate to one
another. However conceptualization also should bring up abstractions
that enable the development of multi-agent systems. Agents can be
implemented with Java or Fortran but these languages are not designed
with agent systems in mind, so they do not offer the right abstractions
for building such systems (although some are very useful). From a
software engineering point of view, we need to map the high level
abstractions deduced from conceptualization to proper technologies.
Language technology is just one example (and maybe an ultimate one).
This seems for me the only way to enable agent technology in practice on
a large scale.
Danny Weyns,
AgentWise, DistriNet
KULeuven - Belgium
> Agree.
>
> Threads are modelling concepts that are quite close to some
> implementations (even to FORTAN ones).
>
> It is a very useful modeling concept because goues beyond what's
> human. There is an antropomorphising trend that equates agency with
> human-like behavior. Agents can go well beyond that into multiple
> threads of consciousness (something that humans cannot do).
>
> Best,
>
> Ricardo
>
> F.Tolman wrote:
>
>> Indeed "living" is the right word. Living is the first step however.
>> "Responsive" is another qualification that can be added to an Agent
>> object. "Social" is yet another qualification that can be
>> implemented. Don't make the error to confuse the concept with the
>> implementation. Agents can be implemented in Fortran if you like. It
>> is the agent concept that makes all the difference. We are now
>> searching for a clean way to model the concept and its different
>> qualifications.
>>
>> Cheers
>>
>> Frits Tolman
>> ----- Original Message -----
>> From: Ricardo Sanz
>> To: Yanqi
>> Cc: modeling@fipa.org
>> Sent: Wednesday, November 19, 2003 10:21 AM
>> Subject: Re: [Modeling] Re: [Modeling] Agent-oriented: why is it
>> necessary?
>>
>>
>> Hi,
>>
>>
>> > That's to say, agent is an extention of object. So,
>> > the importance of agent is in modeling but not in
>> > implementation, is it?
>> >
>> In some sense, the answer is "Yes". An agent is an extension of an
>> object.
>>
>> But the type of extension makes a big deal. I like to think
>> about them as objects with own threads (active objects). That sets
>> the focus
>> not only on modeling but also on implementation.
>>
>> We think about them as "living" objects.
>>
>> Best,
>>
>> Ricardo
>>
>>
>>
>> Yanqi wrote:
>>
>> Hi, Ricardo and all,
>>
>> Jennings wrote in "An agent-based approach for building
>> complex software systems" that no quantitative data is
>> available to demonstrate the efficacy of agent-oriented
>> techniques. He had to use a qualitative justification
>> to argure why agent-oriented approaches are well suited
>> to engineering complex, distributed software system.
>>
>> James Odell wrote "Objects and agents: how do they
>> differ". In the paper, he argues that "agent-based way
>> of thinking brings a useful and important perspective
>> for system development."
>>
>> Is agent just useful in "thinking"? Existing agent-
>> based applications used to model with agents but imple-
>> ment with objects. A common sense of my colleague think:
>>
>> agent = object + mental attitudes
>>
>> That's to say, agent is an extention of object. So,
>> the importance of agent is in modeling but not in
>> implementation, is it?
>>
>> I believe the answer is "No". But I can not make it
>> so clear to my colleagues. :(
>>
>> with regards,
>> qi
>>
>> Quoting Ricardo Sanz <Ricardo.Sanz@etsii.upm.es>:
>>
>> Hi All,
>>
>> It is known that discussions about necessity are
>> always difficult :-)
>> The case for agent-based construction is stronger
>> where there is a need of
>> strong ecapsulation of semiautonomous behavior. The
>> best example I know
>> is a distributed control system (DCS).
>>
>> A control loop is a paradigmatical task for an agent
>> (sense-perceive-act) and
>> being part of a large structure like a DCS forces -by
>> nature, so to say- an
>> agent-based approach.
>>
>> In fact these people were building these type of
>> systems much before the
>> term "agent" was common in mainstream software.
>>
>> Best,
>>
>> Ricardo
>>
>>
>> yan qi wrote:
>>
>> Dear all,
>>
>> After reading lots of papers on agent technology, I
>> have to ask myself this question: is agent-oriented
>> a necessary technology, or as AgentLink says, an
>> enabling technology for next generation software?
>>
>> To answer this question, both academic and practical
>> reasons should be put forward. Yes, we can say agent
>> is autonomous and proactive. But can we show a
>> working example that is can not be well developed
>> by existing software development methodologies?
>>
>> I believe this question is hard to answer in a simple
>> way. Do you think so?
>>
>> I like to listen whatever you say.
>>
>> Best regards,
>> qi
>>
>> =====
>>
>>
>> __________________________________________________
>> Do You Yahoo!?
>> Tired of spam? Yahoo! Mail has the best spam
>> protection around
>> http://mail.yahoo.com
>> _______________________________________________
>> Modeling mailing list
>> Modeling@www.fipa.org
>> http://fipa.org/mailman/listinfo/modeling
>>
>>
>>
>> --
>>
>>
>>
>> ========================< Cut Here
>> =====================
>>
>> Ricardo.Sanz@ETSII.UPM.es
>> Ricardo.Sanz@ASLab.org
>> Ricardo.Sanz@IEEE.org
>>
>> http://aslab.disam.etsii.upm.es/~sanz
>>
>>
>>
>> The AUTONOMOUS SYSTEMS LABORATORY
>>
>> Universidad Polit¨¦cnica de Madrid
>> Madrid, Spain
>>
>> http://www.aslab.org
>> ========================< Cut Here
>> =====================
>>
>>
>>
>> _______________________________________________
>> Modeling mailing list
>> Modeling@www.fipa.org
>> http://fipa.org/mailman/listinfo/modeling
>>
>>
>>
>>
>> Department of Computer Science and Techonology,
>> National University of Defense Techonology,
>> Changsha, 410073 China
>> +86 731 4533854
>> http://yanqi.topcities.com
>> _______________________________________________
>> Modeling mailing list
>> Modeling@www.fipa.org
>> http://fipa.org/mailman/listinfo/modeling
>>
>>
>>
>>
> --
>
>
>
> ========================< Cut Here >=====================
>
> Ricardo.Sanz@ETSII.UPM.es
> Ricardo.Sanz@ASLab.org
> Ricardo.Sanz@IEEE.org
>
> http://aslab.disam.etsii.upm.es/~sanz
>
>
>
> The AUTONOMOUS SYSTEMS LABORATORY
>
> Universidad Politécnica de Madrid
> Madrid, Spain
>
> http://www.aslab.org
> ========================< Cut Here >=====================
>
>
--------------968E56FCC7A35F152AC941E9
Content-Type: text/html; charset=us-ascii
Content-Transfer-Encoding: 7bit
<!doctype html public "-//w3c//dtd html 4.0 transitional//en">
<html>
<body text="#000000" bgcolor="#FFFFFF">
Hello,
<p>Concepts and implementation should not be confused but agents systems
have to be build! So far agent research has mainly focussed on conceptualization.
Conceptualization intends to clarify what the building blocks of multi-agent
systems are and how they relate to one another. However conceptualization
also should bring up abstractions that enable the development of multi-agent
systems. Agents can be implemented with Java or Fortran but these languages
are not designed with agent systems in mind, so they do not offer the right
abstractions for building such systems (although some are very useful).
>From a software engineering point of view, we need to map the high level
abstractions deduced from conceptualization to proper technologies. Language
technology is just one example (and maybe an ultimate one). This seems
for me the only way to enable agent technology in practice on a large scale.
<p>Danny Weyns,
<br>AgentWise, DistriNet
<br>KULeuven - Belgium
<blockquote TYPE=CITE>Agree.
<p>Threads are modelling concepts that are quite close to some implementations
(even to FORTAN ones).
<p>It is a very useful modeling concept because goues beyond what's human.
There is an antropomorphising trend that equates agency with human-like
behavior. Agents can go well beyond that into multiple threads of consciousness
(something that humans cannot do).
<p>Best,
<p>Ricardo
<p>F.Tolman wrote:
<blockquote type="cite"
cite="mid00a001c3aec8$c6d3d710$5d523850@tuqt3gvqlc5bpo">
<pre wrap="">Indeed "living" is the right word. Living is the first step however. "Responsive" is another qualification that can be added to an Agent object. "Social" is yet another qualification that can be implemented. Don't make the error to confuse the concept with the implementation. Agents can be implemented in Fortran if you like. It is the agent concept that makes all the difference. We are now searching for a clean way to model the concept and its different qualifications.
Cheers
Frits Tolman
----- Original Message -----
From: Ricardo Sanz
To: Yanqi
Cc: <a href="mailto:modeling@fipa.org" class="moz-txt-link-abbreviated">modeling@fipa.org</a>
Sent: Wednesday, November 19, 2003 10:21 AM
Subject: Re: [Modeling] Re: [Modeling] Agent-oriented: why is it necessary?
Hi,
</pre>
<blockquote type="cite">
<pre wrap="">That's to say, agent is an extention of object. So,
the importance of agent is in modeling but not in
implementation, is it?
</pre>
</blockquote>
<pre wrap=""><!---->
In some sense, the answer is "Yes". An agent is an extension of an object.
But the type of extension makes a big deal. I like to think
about them as objects with own threads (active objects). That sets the focus
not only on modeling but also on implementation.
We think about them as "living" objects.
Best,
Ricardo
Yanqi wrote:
Hi, Ricardo and all,
Jennings wrote in "An agent-based approach for building
complex software systems" that no quantitative data is
available to demonstrate the efficacy of agent-oriented
techniques. He had to use a qualitative justification
to argure why agent-oriented approaches are well suited
to engineering complex, distributed software system.
James Odell wrote "Objects and agents: how do they
differ". In the paper, he argues that "agent-based way
of thinking brings a useful and important perspective
for system development."
Is agent just useful in "thinking"? Existing agent-
based applications used to model with agents but imple-
ment with objects. A common sense of my colleague think:
agent = object + mental attitudes
That's to say, agent is an extention of object. So,
the importance of agent is in modeling but not in
implementation, is it?
I believe the answer is "No". But I can not make it
so clear to my colleagues. :(
with regards,
qi
Quoting Ricardo Sanz <a href="mailto:Ricardo.Sanz@etsii.upm.es" class="moz-txt-link-rfc2396E"><Ricardo.Sanz@etsii.upm.es></a>:
Hi All,
It is known that discussions about necessity are
always difficult :-)
The case for agent-based construction is stronger
where there is a need of
strong ecapsulation of semiautonomous behavior. The
best example I know
is a distributed control system (DCS).
A control loop is a paradigmatical task for an agent
(sense-perceive-act) and
being part of a large structure like a DCS forces -by
nature, so to say- an
agent-based approach.
In fact these people were building these type of
systems much before the
term "agent" was common in mainstream software.
Best,
Ricardo
yan qi wrote:
Dear all,
After reading lots of papers on agent technology, I
have to ask myself this question: is agent-oriented
a necessary technology, or as AgentLink says, an
enabling technology for next generation software?
To answer this question, both academic and practical
reasons should be put forward. Yes, we can say agent
is autonomous and proactive. But can we show a
working example that is can not be well developed
by existing software development methodologies?
I believe this question is hard to answer in a simple
way. Do you think so?
I like to listen whatever you say.
Best regards,
qi
=====
__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam
protection around
<a href="http://mail.yahoo.com" class="moz-txt-link-freetext">http://mail.yahoo.com</a>
_______________________________________________
Modeling mailing list
<a href="mailto:Modeling@www.fipa.org" class="moz-txt-link-abbreviated">Modeling@www.fipa.org
</a><a href="http://fipa.org/mailman/listinfo/modeling" class="moz-txt-link-freetext">http://fipa.org/mailman/listinfo/modeling
</a>
--
========================< Cut Here
=====================
<a href="mailto:Ricardo.Sanz@ETSII.UPM.es" class="moz-txt-link-abbreviated">Ricardo.Sanz@ETSII.UPM.es</a>
<a href="mailto:Ricardo.Sanz@ASLab.org" class="moz-txt-link-abbreviated">Ricardo.Sanz@ASLab.org</a>
<a href="mailto:Ricardo.Sanz@IEEE.org" class="moz-txt-link-abbreviated">Ricardo.Sanz@IEEE.org
</a><a href="http://aslab.disam.etsii.upm.es/~sanz" class="moz-txt-link-freetext">http://aslab.disam.etsii.upm.es/~sanz</a>
The AUTONOMOUS SYSTEMS LABORATORY
Universidad Polit¨¦cnica de Madrid
Madrid, Spain
<a href="http://www.aslab.org" class="moz-txt-link-freetext">http://www.aslab.org</a>
========================< Cut Here
=====================
_______________________________________________
Modeling mailing list
<a href="mailto:Modeling@www.fipa.org" class="moz-txt-link-abbreviated">Modeling@www.fipa.org
</a><a href="http://fipa.org/mailman/listinfo/modeling" class="moz-txt-link-freetext">http://fipa.org/mailman/listinfo/modeling
</a>
Department of Computer Science and Techonology,
National University of Defense Techonology,
Changsha, 410073 China
+86 731 4533854
<a href="http://yanqi.topcities.com" class="moz-txt-link-freetext">http://yanqi.topcities.com
</a>_______________________________________________
Modeling mailing list
<a href="mailto:Modeling@www.fipa.org" class="moz-txt-link-abbreviated">Modeling@www.fipa.org
</a><a href="http://fipa.org/mailman/listinfo/modeling" class="moz-txt-link-freetext">http://fipa.org/mailman/listinfo/modeling
</a>
</pre>
</blockquote>
<pre class="moz-signature" cols="72">--
========================< Cut Here >=====================
<a href="mailto:Ricardo.Sanz@ETSII.UPM.es" class="moz-txt-link-abbreviated">Ricardo.Sanz@ETSII.UPM.es</a>
<a href="mailto:Ricardo.Sanz@ASLab.org" class="moz-txt-link-abbreviated">Ricardo.Sanz@ASLab.org</a>
<a href="mailto:Ricardo.Sanz@IEEE.org" class="moz-txt-link-abbreviated">Ricardo.Sanz@IEEE.org
</a><a href="http://aslab.disam.etsii.upm.es/~sanz" class="moz-txt-link-freetext">http://aslab.disam.etsii.upm.es/~sanz</a>
The AUTONOMOUS SYSTEMS LABORATORY
Universidad Politécnica de Madrid
Madrid, Spain
<a href="http://www.aslab.org" class="moz-txt-link-freetext">http://www.aslab.org</a>
========================< Cut Here >=====================
</pre>
</blockquote>
</body>
</html>
--------------968E56FCC7A35F152AC941E9--