An Event Detection Approach for Identifying Learning Evidence in Collaborative Virtual Environments

3D virtual environments support collaborative learning through connecting users in real-time allowing them to accomplish learning tasks together, in addition they enhance the students' exploration, engagement, and interactivity. However, collecting learning evidence to evaluate students in these environments has many difficulties. Therefore, the intention of this paper is to describe our approach for assessing student's learning within collaborative groups in 3D virtual worlds (VWs). It combines a computational mechanism that integrates software agents and natural agents (users) with an ontology approach that supports the identification of learning evidence from collaborative activities that mimics classroom observation. The software agents track the users and collects different actions, clicks, and events to evaluate the quantity of interactions, while the natural agents perform peer evaluations of the students to assess the quality of their performance. The aim is that such a computational model can support more in-depth assessment of learning activities in 3D spaces.