Using Misunderstanding and Discussion in Dialog as a Knowledge Acquisition or Enhancement Procecss - Distributed Artificial Intelligence, Agent Technology, and Collaborative Applications - page 88

in our work, appears when the student cannot attach the teacher's knowledge to is KB, because it lacks
concepts, or it lacks implications. Discussion appears with two types of profiles: Students with wrong
knowledge, and curious students (those who want to learn more than what is in their lesson). We have
presented an example of misunderstanding as well as a discussion sample dealing with knowledge conflict
management. While scrutinizing both situations we noticed a symmetry between the dialogic structure
and the KB management process. Since they were seen as repair procedures through which the teacher
helps the student restore its KB integrity and connexity and proceed further in the lesson, they were both
subsumed into a generic pattern containing four items: A trigger (which causes the need for repair), a
repair process launched either by the teacher (in misunderstanding) or by the student (in discussion) in
which the main actor directs the dialog, a repairing action (with a given FR: give - explanation () in the
case of misunderstanding, and give - information ( F alse ) in the case of a knowledge falsification needing
discussion), and last an evaluation of the repair , uttered by the student.
As much satisfying as it can been, the socratic lesson is a particular case of a wide range of knowl-
edge acquisition and revision using communication. This is why we think that some of our results, here
isolated in a sort of in vitro state, could be, in turn, experimented in more flexible environments. For
instance, agents might learn from each other (symmetrical role) after acknowledging each other's skills,
which will require a serious enhancement of our simple 'other agent's model' embedded in the system
architecture. Agents might also check knowledge with another knowledge source, which will provide
contradiction, and possibly an interesting set of discussion procedures, that might involve more than a
pair, and so forth. Nevertheless, we think that this study has shown that the assumption of symmetry
between dialog and knowledge management is a plausible one, and most of all, an implementable solu-
tion, and nothing for the moment prevents its extension to more complex situations.
r eferences
Allen, J., & Perrault, R. (1980). Analyzing intention in utterances. Artificial Intelligence , 15(3) , 11-18.
Amgoud, L., & Prade, H. (2003). A possibilistic logic modeling of autonomous agents negotiation. In
Epia (p. 360-365).
Angluin, D. (1987). Queries and concept learning. Machine Learning , 2 (4), 319-342.
Asoh, H., Motomura, Y., Hara, I., Akaho, S., Hayamizu, S., & Matsui, T. (1996). Acquiring a probabi-
listic map with dialogue-based learning. In H. Hexmoor and L. Meeden (Eds.), ROBOLEARN '96: An
International Workshop on Learning for Autonomous Robots , (pp. 11-18).
Baker, M. (1994). A model for negotiation in teaching-learning dialogues. Journal of Artificial Intel-
ligence in Education , 5 (2), 199-254.
Beun, R.-J., & van Eijk, R. M. (2003). A cooperative dialogue game for resolving ontological discrepancies.
In Workshop on agent communication languages. (Lecture notes in computer science, pp. 349-363).
Beun, R.-J., & van Eijk, R. M. (2005). Conceptual mismatches and repair in human-computer interac-
tion. In Proceedings of the Seventh Belgium-Netherlands Conference on Artificial Intelligence, BNAIC,
(pp. 315-316).