Towards the Application of Argumentation-based Dialogues for Education. Elizabeth Sklar Department of Computer Science Columbia University New York, NY 10027, USA sklar@cs.columbia.edu Simon Parsons Department of Computer and Information Science Brooklyn College, City University of New York Brooklyn, NY 11210, USA parsons@sci.brooklyn.cuny.edu Abstract This paper describes our work constructing a gen- eral framework for modeling multi agent interactions in education-related applications. We are motivated to move beyond a traditional scripted model and fol- low the general trend in human education towards more open, learner-centered, constructivist environments. In or- der to accomplish this, we need a framework in which to define general types of interactions that can occur be- tween a learner and a tutor, as well as interactions be- tween these agents and their sets of beliefs — not only about the knowledge domain that is the subject of the learn- ing system, but also about each other. In this paper, we describe early work in this direction, which involves us- ing argumentation and extending existing dialogue pro- tocols to allow for various types of tutor-learner interac- tions. 1. Introduction We are interested in constructing a general frame- work for modeling multi agent interactions in education- related applications. Currently, we are working on two such projects. One is traditional in the sense that it con- cerns building agents to interact with human learners in a web-based interactive learning environment [8]. The other involves modeling the education system as a multi agent simulation in order to be able to demonstrate and ex- plore the types of interactions and interplays that occur between teachers and students in classrooms, princi- pals and teachers in schools, superintendents and teachers in school districts, and so on [9]. Historically, interactive learning systems (ILS), in gen- eral, and intelligent tutoring systems (ITS), more specifi- cally, are highly engineered to the particular knowledge do- main to which they are applied e.g., [1]. We are motivated to explore a more general methodology for interactions. Fol- lowing the direction of education over the last 30 years, there is a general trend towards learner-centered learning, where the learner takes the initiative and the teacher (or tutoring system) offers support but not the same kind of teacher-centered instruction that had been used previously [2, 3, 5, 7]. In a learner-centered environment, the learner actively takes the initiative in structuring his/her own learn- ing. With this learner-centered trend in mind, we are work- ing towards building on-line learning environments that cannot be scripted because, by definition, the direction of the learning comes from the student and cannot be engi- neered a priori. In order to accomplish this, we need a framework in which to define general types of interactions that can oc- cur between a learner and a tutor, as well as interactions be- tween these agents and their sets of beliefs — not only about the knowledge domain that is the subject of the learning sys- tem, but also about each other. In this paper, we describe early work in this direction. Here we give an explanation of the interaction models we are using, and the full paper de- scribes the pre and post conditions of each type of model, and details the changes in the belief sets of both types of agents that they entail. We have chosen argumentation as our interaction model. Argumentation-based dialogues, often specified as dialogue games allow agents to engage in “conversation” for a va- riety of purposes and enable systems to reach beyond re- source allocation tasks [4], which are what commonly used approaches to agent interaction like auctions and negotia- tion were designed to address. 2. Education dialogues Dialogues for education take place between two agents, each having specific roles. In a traditional classroom, these could be considered a teacher and a student. Here, we refer to these agents more generally as T utor and Learner. This allows us the ability to apply the dialogic framework (de- Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. AAMAS'04, July 19-23, 2004, New York, New York, USA. Copyright 2004 ACM 1-58113-864-4/04/0007...$5.00