Presentation of how we implemented the reasoning module according to the design

1. Contents What we have done Architecture Reasoning strategy Conclusion Future work
Implementation of the Reasoning Module
Damien Clauzel
American University of Paris
Project meeting in Paris, 21-22 May 2007

2. Contents What we have done Architecture Reasoning strategy Conclusion Future work
Contents
What we have done
Goals
Our work environment
Architecture
Detailing the architecture
Blackboard
Events agents
Integration & intervention agents
Reasoning strategy
Conclusion
Future work

3. Contents What we have done Architecture Reasoning strategy Conclusion Future work
What we have done
We wanted to achieve the following goals:
• implement the Reasoning Module according to the design
• support targeted scenarios from the conceptual framework
(3,4 and 7)
• connect the Reasoning Module with other components from
the project
Conceptual Framework (D1.3) → Design (D2.2) → Prototype (D3.2)

4. Contents What we have done Architecture Reasoning strategy Conclusion Future work
Our work environment
Focusing on open standards for interoperability and integration in
heterogeneous environments:
• Java application server and applet
• mySQL database server
• web services based communication
⇒ Only open source components

5. Contents What we have done Architecture Reasoning strategy Conclusion Future work
Design architecture

6. Contents What we have done Architecture Reasoning strategy Conclusion Future work
Implementation architecture
Communication
Application agents intervention agent
interventions
SOAP
Blackboard
ASKME Web services
events foci integration agent
User Event
Task Foci
idleUser Continuation NewInfoAvble Start
intervention agent agent agent agent
Databases Reasoning
with models Database interfaces

7. Contents What we have done Architecture Reasoning strategy Conclusion Future work
Blackboard
A blackboard based implementation
• used for all communication between agents
• collect facts about the user and his activity
• 1 learner = 1 blackboard

8. Contents What we have done Architecture Reasoning strategy Conclusion Future work
Event agents
Agents computing the possible targets of attention
• in charge of creating possible foci from events
• autonomous
• of various type (target an event or a concept)

9. Contents What we have done Architecture Reasoning strategy Conclusion Future work
Integration & intervention agents
Integration agent: refine the foci from the event agents
⇒ clean, merge and optimize
Intervention agent: prepare and manage the interventions
⇒ uses breakpoints

10. Contents What we have done Architecture Reasoning strategy Conclusion Future work
Reasoning strategy
Some reasoning and responses can be generalized:
• default set of rules for supporting common generic situations
• expandable by the hosting application
• from an event, generate several interventions for the
application to choose from
Example: User becomes idle (IdleInput event)
⇒ attention management: re-attract user attention
⇒ cognitive support: propose help for the task

11. Contents What we have done Architecture Reasoning strategy Conclusion Future work
Conclusion
• prototype is working
• development points to a lot of things to improve
• require testing and validation

12. Contents What we have done Architecture Reasoning strategy Conclusion Future work
Future work
• provide high level macro language for the rules
• implement more scenarios from the conceptual framework
• support meta-reasoning