CSCL Activities

CSCL Workshop

Meet us at ISLS! We are organizing this year's CSCL workshop to build a CSCL vision together. This workshop welcomes scholars of all backgrounds and stages of their CSCL journey to meet and exchange around a vision for the field, In a unique approach, we intend to actively and collectively shape the future of the field. As researchers passionate about collaborative learning, we consider this endeavour a learning activity itself in which we rely on everyone in the community to build the future of CSCL.  Find out more here

Conceptual Foundations of CSCL

The workshop is an actionnable we identified after conducting a survey within the CSCL community about its member's perceptions of key concepts in the field and future challenges in the field. This very insightful study will be presented at ISLS 2025 in Helsinki, Finland. Details about the presentation schedule will be posted here soon.

This study is the result of our very first collaboration. Since we have had such a great time doing research together in this adhoc collaboration, we are convinced that others can enjoy this and learn from it  just as much as we did. We thus plan to propose another fully open collaborative study likely for the 2nd semester 2025. Stay tuned!

Generative AI for Qualitative Research

Multi-Agent Systems for Thematic Analysis

We have developed a novel multi-agent-system based on the Amazon Bedrock infrastructure able to conduct inductive thematic analysis. The system produces a complete report with negotiated codes, themes and descriptions. 

We have tested the system with our CSCL survey dataset where participants were asked to provide definitions for CSCL concepts - data that no language model has been trained on and with definitions oftentimes only different in nuances. The results of the system have been deemed "impressive" by the manual coders - read the complete paper here

Methods for Analyzing the Quality of Automated Thematic Analysis Systems

With the advent of fully automated system providing the ability of thematic analysis, we now can ask if and how we can compare the outputs on a same dataset. Braun and Clarke (2019) advocate against considering the data holding universal truth in a positivist sense, and we agree! Still, we believe that the creativity and perspective taking of different teams and systems can be assessed - similar to intercoder-reliabity and similar measures! A preprint for our submission to AIED will be available soon.