Towards a model of learning via dialogue

Reviewer: Jose M. Ramirez

This paper addresses the problem of learning from the perspective of the communication between agents. Obeid covers understanding as an essential element in the process of learning. The author briefly treats several problems related to information exchange—consistency, monotonicity, interaction, and consensus—in a logical sequence that leads to the idea of a computational model based on a propositional modal nonmonotonic logic called LPR1 that has two modal notions: M for Plausibility and P for Possibility. The model is very simple, but the definitions correctly handle both the exchange of beliefs about a particular domain and the structure of the information exchanges. The model's mechanism for screening inappropriate interpretations is weak. The model can be applied to some information exchange processes, such as knowledge acquisition, including the concepts of beliefs and common sense. The paper is well written and fairly self-contained; Obeid briefly discusses the problems that motivated the study. The references are very useful.