Using Misunderstanding and Discussion in Dialog as a Knowledge Acquisition or Enhancement Procecss - Distributed Artificial Intelligence, Agent Technology, and Collaborative Applications - page 86

Figure 7. The tutored learning system through dialog between artificial cognitive agents
not correct evolution of the situation is observed). The 'World' is a sharable knowledge base, a pool of
predicates available to agents.
Each agent has goals, a KB and model of its interlocutor. A model of the interlocutor is what the
agent knows that its interlocutor knows. The 'teacher' agent mostly checks its interlocutor's model
and updates it, by asking the 'student' questions, when it (the 'teacher') needs to explain something
to the 'student'. The 'teacher' also knows what the 'student's' goals are in order to be prepared to any
discussion shift. For instance, when the 'student' tries to fulfill its curiosity by asking some knowledge
and thus breaking the lesson flow, the 'teacher' will be able to understand the 'student's' intervention
and satisfy its request before continuing the lesson. The model the 'student' has of its 'teacher' is that
all what the 'teacher' says is true, and thus a give - knowledge ( p ) clause is equivalent to P is True (irst
axiom). It also have an idea of the goals and some knowledge of its 'teacher'.
Naturally, each agent when shifting from a role to another, in a different situation, modifies its
interlocutor's model according to its present role. It can freely update them in order to make them evolve.
Each agent action is motivated by its own goals, which lead the agent to use the adequate strategies
and FR rules.
Implementation
We have implemented a Java program to test conflict solving. This program is a basic prototype aiming
at getting experimental results of a part of our theory. We describe this prototype in (Yousfi-Monod
& Prince, 2007). The running of our prototype program has shown that the 'student' has been able to
detect a conflict between its KB and a new data provided by the 'teacher' and then ask the 'teacher' to
validate some potentially conflictual knowledge and finally remove the wrong implication.