"Affective Computing and the use of Pedagogical Agents in E-learning"

E-learning technologies, recognized today as a valuable tool to support both instructors and learners, have developed into a learner-centered, dynamic and personalized process. In comparison to face-to-face instruction in class, however, e-learning is still considered impersonal and deficient in terms of communicating the emotional of affective state of the learner, due to the human-computer interaction limitations posed by contemporary technology. Affective computing (AC) aims to enhance human-like aspects in e-learning, by analyzing, recognizing and interpreting the affective state of human learners. AC is realized through systems that register human reactions in real time, while the learner interacts with the digital learning platform and content. In this framework, the present thesis aims to investigate the following:

• Automatic classification/recognition of the learner’s affective state on the basis of a model for affects that is suited to e-learning activities.
• Choice and development of a pedagogical agent (preferably animated) suited to the specific class of learners, who will offer timely aid to the learner and convert his/her affective state to the positive. Successful incorporation of the agent in an e-learning environment is expected to prevent phenomena like boredom, loss of motivation and interest in the learning process, reduced participation and eventual drop out, or poor evaluation results.

Development of an automated, integrated educational model that incorporates the pedagogical agent functionalities to keep learned engaged and active in the learning process, and to ensure his/her feeling of satisfaction from the overall e-learning experience upon completion of it.