Automatic Support for the Analysis of Online Collaborative Learning Chat Conversations

The paper starts from experiments of using online instant messenger chats as an assignment in a classical face to face course. Tutors reported that the activity of evaluating and grading students' chats is very difficult. Starting from the analysis of their evaluations, Bakhtin's polyphonic theory was chosen as a good starting point for developing automatic support tools. As a consequence, the PolyCAFe system based on this theory was developed using techniques from Natural Language Processing and Social Network Analysis. The system provides interactive graphics visualization of the chat and textual feedback. The persons who will beneficiate the most from the system are tutors. They can use the system for analyzing the participation of students and getting insights for writing a detailed feedback and grading. Students also may use the system, for getting feedback.