Multi-Agent Systems in a Computational Environment of Education: A Chatterbot Case Study André F. M. Batista, Maria G. B. Marietto, Gislene C. O. Barbosa, Robson S. França and Emerson A. Noronha Federal University of ABC – Center of Mathematics, Computation and Cognition Santo André, São Paulo - Brazil {andre.batista, graca.marietto, gislene.barbosa, emerson.noronha}@ufabc.edu.br Abstract This paper proposes a dialogue based interface for virtual environments of learning with a Multi- agent approach. Here, it's shown both aspects involved with the motivation and conception of this interface project, the conceptual and the technological aspect. In the conceptual aspect, the concepts related to distance learning and learning process are presented in the context of interfaces for virtual environment for education. In the technological aspect, it's presented the multi-agent based systems, chatterbots systems and the AIML language applied in the context of the conceptual needs. Finally, based on these concepts and technologies, the paper’s contribution is a project for a virtual environment of learning, a middleware architecture based on a multi-agent approach. 1. Introduction The ability to apply computers in the teaching- learning process by creating educational virtual environments provides interactive situations which the student may arise as an active player in his knowledge construction. Lévy [13] highlights the impact of computers in the individuals’ cognitive skills by the interface features, the software itself and its tools. Considering the graphic interface alone, the most part of the interface is based on hypertext navigation systems and/or taking choices by buttons and menus. Both approaches demand an additional cognitive effort by students because the usual form of communication among humans is the natural language. Therefore, the most suitable cognitive interfaces are those which allow the establishment of dialogs in natural language. This makes the interaction student-system more appealing, and with a higher level of empathy. Among others, the software that has this interface feature is a chatterbot system, and its main goal is to get acquainted with people [8]. In this context, this paper goes on investigating the usage of chatterbots as an interface for education in a virtual environment. Specifically, the application of a chatterbot as an interface for clarifying students' doubts about the Java programming language. This interface based on a chatterbot system will be a supporting tool for students in the content of their classes. The usage of virtual environments in education can contribute in the process of teaching-learning because: (i) it has a more suitable illustrational power on some process and objects, comparing to other types of media; (ii) it allows the student to develop the work at his pace; (iii) it demands more interaction, encouraging the student for a participative role in the learning process. However, without the interpersonal relationship with teachers and classmates in this kind of environment, the student tends to fell discouraged by the artificiality of the human-machine interaction, resulting in loss of interest of the educational process [17]. The application of chatterbots as virtual assistants to simulate a dialog with partners can stimulate students, decreasing their feeling of isolation in these environments. Additionally, these kind of systems provide a feeling of security and satisfaction because they are always accessible, providing to the student an always available virtual tutor to chat and/or to settle his doubts. The following works are examples of chatterbots applied in education: Júnior [13;14], Electra [9], BonoBOT [15]. This paper is organized as follows. Initially, Section 2 presents a general view of multi-agent based systems. Next, in Section 3, the concepts related to chatterbots are presented, along with a description of the AIML language, used in chatterbots (Section 3.1). After that, Section 4 presents the concepts and motivations related to the usage of chatterbots in distance learning and it is followed by the requirements and the proposed software architecture. Finally, Section 4.2 presents the construction of the knowledge base, and this project’s mail goal is later illustrated in Section 4.3 in a practical manner. 2. Multi-agent Based Systems The Distributed Artificial Intelligence (DAI) has emerged from the integration of the Artificial Intelligence (AI) area and Distributed Systems (DS) in the 80s, with theoretical and practical implications from both areas. With cognitive and social theories International Journal for Infonomics (IJI), Volume 3, Issue 3, September 2010 Copyright © 2010, Infonomics Society 285