Enhancing social and collaborative learning using a stacked GNN-based community detection

Social and collaborative learning (SCL) aims to provide an inclusive and interactive learning experience that goes beyond traditional classroom settings. By leveraging social learning analytics in SCL, educators can create dynamic and engaging learning experiences that empower students to work within effective communities. To do so, SCL must take advantage of technological innovations of AI to build innovative digital educational solutions for the detection of communities. In this context, this research aims to identify some of the most promising trends for building effective SCL in higher education context, the capabilities offered by Social Networks Analytics based on Graph Neural Networks (GNN). Findings highlight a stacked GNN-based ensemble clustering approach of community detection which provides complementary understandings of best practices when aligning social learning goals and theories with technological capabilities. In fact, our proposal argues that the definition of SCL should be built around socio-constructivism theory. Additionally, it is based first on an ensemble clustering that seeks to form effective communities i.e. homogenous groups of learners basing on their social interactions and a transfer learning to improve community detection performance. To validate the proposed approach and to assess its technical viability, we quantitatively rely on evaluation metrics that demonstrated the performance of the proposal, and we qualitatively introduce a real case study where students have shown their satisfaction within a real SCL setting at a higher education level.