ELAI: intelligent agent adaptive to the level of expertise of students

Abstract

The use of intelligent agents aware of the individual characteristics and context of students, allows us to offer a suitable support to the real requirements. Allied to this, the implementation of these agents in the three-dimensional immersive virtual environments (3D), tend to transcend the existing potential in the interactions with the learning objects it contains and also to expand the alternatives of building the students’ knowledge. In this way, throughout this paper it will be presented the development of intelligent agent called The Intelligent Agent adaptive to the Level of Expertise of Students, by using the Non Player Character (NPC) strategy existing on the virtual world platform OpenSimulator. The ELAI provides support for teaching Computer Networking, being sensitive to the context of learners, regarding to their level of knowledge (expertise). In order to maximize the flexibility of interactions between the student, through the student’s avatar and the NPC, an interconnection of this agent with a chatterbot was established, whose knowledge base was increased by classes in AIML inherent to the topic of Computer Networks. As a result, tests were conducted with students of the Federal University of Santa Maria, participating students of undergraduate programs in Computer Science, Information Systems, and Post-Graduate in Computer Science at Masters Level, where they were instructed to follow an Educational journey in order to maintain a uniform investigation. Later, the students were conducted to answer three questionnaires, namely: the System Usability Scale (SUS), which received the usability of intelligent agent implemented in immersive virtual environment; in the second evaluation, it was used the intelligent agents criteria; and in the third evaluation it was asked to students in general about the intelligent agent and the adjustments provided for it. The results obtained during the evaluations show that the intelligent agent ELAI met to the expectations of the evaluators, analyzing that it was acquired the classification of "73.04" the in SUS usability testing range, which it is equivalent of "Good" in the classification and average score method. In the other two questionnaires it was administered the Cronbach’s alpha coefficient to estimate reliability of students’ responses, which obtained the value of "1.08" in the first questionnaire and the second the value of "1.06", resulting to "Excellent" in the Cronbach’s alpha coefficient ranking.