Multi-Agent Architectures Supporting Distributed Learning
in a Wired and Wireless Future

May 19, 2001


associated with

San Antonio, May 19-23, 2001

(and held in close timely and geographical proximity with Agents'2001, Montreal, May 28-June 1, 2001)

NEW!!! Workshop Programme

On-Line Proceedings (WinZipped Postscript, 3.25 MB)


The future learning environments will be wired and wireless, accessible from anywhere at anytime. Learning in these environments will be distributed in space and time. Standard classroom models for knowledge building will be complemented with virtual classroom models involving people of different ages, cultural and knowledge backgrounds. Knowledge building will be a lifelong, social, evolutionary process of building consensus through sharing and discussing (knowledge negotiation). How should environments supporting this type of learning be built? Several types of techniques and technologies hold a promise: multi-agent architectures, user and learner modelling, mobile and ubiquitous technologies.

Areas of Interest

Multi-agent architectures are based on software agents, autonomous software components, that can interact through a standard protocol and collaborate with each other to achieve common gaols. Such architectures are promising, since they are inherently distributed, modular and open. Through the uniformity of agents and through a standardized interaction protocol, a level of scalability and interoperation can be achieved that is impossible with other techniques. Such architectures allow for additiveness and heterogeneity in the software environment. However, currently there aren't many distributed learning environments based on large - scale mulit-agent system architectures. Why is this the case? What are the advantages and disadvantages of existing agent-development environments? What are the difficulties underlying the development of such environments? How can one ensure real scalability, how can one cope with the system complexity and the unpredictability in its behaviour? The workshop will focus on some of these difficulties and stimulate questions and answers from participants with experience in developing multi-agent based environments.


Another potentially useful class of techniques come from AI. Personalization of the environment (selection of "community" or "virtual classroom", translation and dialogue adaptation, adaptation of presentations / discussions to the individual level of knowledge, preferences, tasks and learning style) can be achieved through learner modelling. Which techniques in particular are potentially useful? Which techniques are lightweight enough to come on board the agents? How can such techniques be incorporated in a multi-agent environment? Where is the place for these techniques: in the individual agent, in agents playing the role of tutors or artificial learners (learning companions), or in a centralized, omnipotent facilitator-component?


Mobile computing devices are rapidly gaining power and connectedness through hardware technologies like BlueTooth and software technologies like Wireless Ethernet, Cellular Digital Packet Data (CDPD) and Wireless Application Environment (WAE). The growth in this area is so fast that every state of the art report is already outdated at the moment it is written. Mobile technologies already allow running powerful applications, downloaded dynamically as the carrier of the device enters the range of a station. With such devices and the emerging ubiquitous computing environments we can imagine changes in the classroom very soon. For example, the students won't need to carry heavy-loaded backpacks to school anymore, but will bring only their palmtop computer, which will interact with the school desk (a large horizontal touch screen with embedded computer), and download all homework, projects and materials needed while in class. We can only speculate how these devices could be usefully applied for learning out of the classroom. One of the goals for the workshop will be to discuss these ways.


Finally, distributed learning environments raise many educational and sociological issues. Who defines the "curriculum" in such a democratic society of learners? If learning emerges from the interactions in the virtual classroom, should it be and can it be targeted, or driven in some particular direction? How can a fruitful knowledge building interaction be ensured? What would be the role of the teacher - facilitator, or something else? What would be the role of the environment -communication medium only, or an active participant in the learning process? What is an appropriate metaphor for the underlying multi-agent system -- ecological: "survival of the fittest (best serving) agents", economical: "agents striving to earn virtual currency", or sociological: "adhering to certain liberal norms of behaviour"? How can one ensure that the resulting behaviour of the system is not only robust and dependable, but that it actually stimulates and motivates learner participation and learning?


Format of the session


The workshop will take one full day. Presentations of the selected papers will take the first part of the workshop. The goal of the presentations will be to describe work in the area, but also to stimulate discussion along the workshop issues. The second part of the workshop will start with a short brainstorming part where working questions will be agreed. Then the workshop participants will split in 4 working groups. Each group will have to brainstorm, discuss and come with answers to their questions. After a one-hour period, the groups will report the results of their discussions and a general discussion and a brief conclusion session will follow.




The workshop will enable participants from different backgrounds and perspectives to share their views and learn from each other about the techniques that will underlie the personalized learning environments of the future. Therefore we invite participants interested (not exclusively) in the following areas:

Submissions in the form of

indicating a particular area or question for discussion should be sent to to Julita Vassileva at as e-mail attachments in any of the following formats: MS Word, Postscript, PDF or RTF.

The committee will review the papers and several of them (maximum 10) will be selected for presentation at the workshop and inclusion in the proceedings. A possible future publication as a peer-edited collection or a special issue of IJAIED will be discussed at the workshop. The authors of selected contributions that were not recommended for inclusion in the proceedings will be also invited to give short presentations and participate in the discussion.


March 15, 2001 Submissions

April 5, 2001 Acceptance / Rejection notification

April 25, 2001 Full papers due (5 pages)


Program Committee:

Liliana Ardissono, University of Torino, Italy

Gary Boyd, Concordia University, Canada

Peter Brusilovsky, University of Pittsburgh, USA

Tak-Wai Chan , National Central University, Taiwan

Robin Cohen, University of Waterloo, Canada

David Franklin, Northwestern University, USA

Ulrich Hoppe, University of Duisburg, Germany

Gord McCalla, University of Saskatchewan, Canada

Joerg Mueller, Siemens AG, Germany

Ana Paiva, Technical University of Lisbon, Portugal

Vittorio Scarano, University of Salerno, Italy

Wouter van Joolingen, University of Amsterdam, Netherlands

Julita Vassileva, University of Saskatchewan, Canada, Chair

Thomas Wagner, University of Maine, USA