Multi-Agent-Based Peer Tutoring in Virtual Learning Environments

Peer tutoring is a collaborative learning approach in which the tutor is attending to a tutee in a one-on-one teaching process. Since education is increasingly moving online, we propose a multi-agent system that dynamically creates tutor-tutee pairs among learners in a virtual learning environment. Our multi-agent system represents learners and instructors. While the instructor agent creates and broadcasts new exercises, the learner agent sequences exercises, evaluates incoming tutoring requests based on their utility value, inserts tutoring requests into the learner's sequence, allows the learner to initiate tutoring requests during an exercise, and allocates these tutoring requests among the other learner agents in the virtual learning environment. Learner agents use the Contract Net Protocol for the allocation of tutoring requests. The inner workings of the learner agent are modeled as a Finite State Machine. Our system is implemented with the JADE framework and demonstrated using a small-scale example.