Introduction

---

The recent increase in information production and the consequent necessity to process it have been justifying research on the use of multiagent systems. The control of power or chemical plants, data searching on the Web, and the study of artificial life are examples of activities that can benefit from the use of these systems. The use of intelligent agents is a useful approach, because they can process huge amounts of data (from sensors or databases), intervene directly in processes, and interact with other agents to achieve their tasks. The aim of this chapter is to discuss the role of agents not only in the enhancement of existing processes but also as a framework with which to design new processes. The chapter focuses on using agents to implement a learning environment that enables its human users to develop social competences rather than just technical ones.

Most learning management systems (LMSs) for either face-to-face or distance education take into account only punctual elements (exercises, quizzes, tests, essays) to produce images of the process that helps educators control these systems and assess their participants. The reason for this limitation is easily understood. Traditional educational models are responding to the social pressure and costs associated with educating masses of people. Classrooms following the face-to-face model often have classes with 30 or 40 students and one educator.

Independently of any pedagogical approach, everyone who has prepared a course has also considered how to assess learning. Which points should be evaluated? How long should students spend on quizzes and exams? How long can teachers spend on analyzing data from quizzes and exams? What is the maximum number of students a teacher can teach effectively? Such constraints have been forcing the focus of assessment to be on content-related competences like memory, concept relations, and the use of models.

Even if they are based on a social philosophical approach, most pedagogical practices are limited to lectures and group work that usually divide the course material into separate tasks and then conclude with a final synthesis. Most of us grew up in such an educational environment, and we can have difficulty accepting that our educations could have been better than they were, especially if we are recognized as skilled professionals. However, our information society is pushing the standards of professional competence beyond what they used to be. It is no longer good enough to master specific technical skills or competences. In our dynamic world, we must also master competences that allow us to collaborate easily on teams, learn new subjects, and adapt ourselves to new working conditions. We call these competences social competences in opposition to professional competences (i.e., those that are developed in traditional education). Nowadays, educational models cannot ensure equality to all students (concerning the development of social competences), because educators have to plan their pedagogical practices by thinking in terms of mass education and content assessment. Even if educators had the ability to do so, they would not have enough time to observe, register, and assess the huge amount of data inherent in social-learning processes.

When shifting from face-to-face to distance-education LMSs, the problem intensifies. The performance indicators they provide reproduce the ones already used in face-to-face environments. Examples are students’ accesses, log-in records, quizzes, and tests.

We believe that the use of intelligent agents applied to online learning environments can enable the design of “enhanced-learning environments” that allow for the development and the assessment of social competences as well as the common professional competences. Examples of social competences include presenting ideas in a workgroup, providing and receiving criticism, cooperating with others, and behaving ethically in one’s working life.

One could argue that online environments can hardly reproduce the richness of real face-to-face interaction. This argument is true if we associate the use of such environments with reproducing traditional human learning scenarios (e.g., classrooms, libraries, lectures). However, these environments can easily create learning scenarios with new roles and rules for human interaction. In order to stimulate creativity and interaction among students, such environments should adopt an exploratory rather than a directive approach to content. Thus, project-based learning (PBL) seems to be a better learning strategy to consider than the traditional strategies. Traditionally, PBL is considered difficult to implement and manage in groups of more than 10 students. Also, the subjective component in how students are evaluated in PBL is a sensitive point of discussion. Both arguments are based on the fact that traditional data describing students’ contributions (frequency, exams, grades) are insufficient when assessing the extent of students’ performances.

We consider the use of intelligent agents as being a good approach for building collaborative online learning environments based on PBL, because these agents can collect huge amounts of data regarding students’ interactions and present these data in a way that allows students and teachers to visualize what is going on and plan what to do. Students can plan their contributions for the projects in which they are participating, and educators can plan how to conduct the learning processes.

“Agents” can be perceived as computing services that humans, or even other agents, can request in order to accomplish their tasks. Some services may be simple and others rather complex. A way to determine the best agents (services) to be implemented is to identify who the actors are in the object of study, which roles they play, and (if possible) what kind of knowledge they use.

Thus, when designing such an environment, the developers should consider the agents as integrating three kinds of services:

1. Helping people to perform innovative activities (i.e., educators need to create groups, projects, assessment portfolios; students have to relate the solutions they create to the problems proposed, to negotiate with other students, to collaborate with them, and to criticize or judge their peers’ work)

2. Stimulating social behavior within students (i.e., if the system determines that two students are working on similar issues, it can inform the students and give them information about how to contact each other)

3. Offering the educators clear and objective information about the students’ performances (i.e., which students are more creative, who effectively produces what, which students cannot collaborate, which students have to improve their reasoning skills)

This chapter introduces a COLE project on which the authors have been working. The basis and architecture of the COLE are explained. In order to facilitate the implementation of particular agents, a generic agent (GAg) and its functionalities are presented.

---