Enhanced E-Learning Community Detection Using Distributed Overlapping Feature Selection-Based Evolutionary Propagation Recurrent Neural Network

Community Detection (CD) is an important aspect of identifying new groups and relations in E-learning Online Social Networks (OSNs). Community identification is a challenging task due to detecting small or overlapping communities having higher resolution future limit problems which is time-consuming and not suitable for large-scale datasets. To alleviate the limitations of previous methods, this paper presents a Community Enhancement Structure-based Evolutionary Propagation Recurrent Neural Network (CES-EPRNN) algorithm. Initially, the Box-Cox Transformation (BCT) algorithm is used to reduce outliers from the education community dataset in OSN. In addition, The Correlation Exhaustive Feature Selection (CEFS) algorithm proficiently identifies the attributes of the education community. Following that, the proposed CES-EPRNN algorithm determines who belongs to the same academic community in OSN. The extensive experiments demonstrate that our proposed technique outperforms conventional methods on the Education Community Dataset (ECD).