Learning Community Detection and Evaluation

Abstract

Community detection in social networks is an interesting topic that attracts the attention of several researchers in the last few years. In this context, several methods that adopt different modelling choices have been proposed in the literature; modularity optimization, agglomerative and divisive techniques, and random walks models are the prominent examples. However, these methods are used in several fields without taking into consideration the type of communities discovered. In the e-learning field, community detection presents a valid tool to analyze and evaluate learners’ tasks. In this study, we are interested in detecting learning communities in social learning environments in order to evaluate learners within their clusters. In this way, we proposed a new based-clique method to detect and evaluate learning communities. Our approach mainly consists of two phases. The first phase is developed to discover the learning community using the maximal clique concept, while the second one is devoted to learning community evaluation based on the interactions among learners and their socio-economic characteristics. In order to test the performance of our method, we use a high-resolution dataset that describes the social interactions among children in a primary school in Lyon, France. In terms of quality performance, our method has proven its efficiency compared to other algorithms.