A New Approach to Detect At-Risk Learning Communities in Social Networks

Abstract

Learning community detection in social networks has an important role to understand and to analyze the network structure. The main objective of our study is to evaluate learning communities based on interactions between learners. To meet this goal, we propose a new algorithm called Community Detection and Evaluation Algorithm (EDCA). This algorithm detects learning communities using a new centrality measure named “safely centrality” that allows to detect safe learners. These learners represent the initial nodes of communities. Afterward, we identify neighbors of each safe learner to build communities. In order to test the performance of our method, we compare our proposed algorithm with three community detection algorithms in two learning networks using the modularity and the Adjusted Rand Index (ARI) metrics. Our experimental phase demonstrates the quality of our proposed algorithm.