Graph Convolutional Neural Networks for Link Prediction in Social Networks

Abstract

Social networks are complex systems that require specialized techniques to analyze and understand their structure and dynamics. One important task in social network analysis is link prediction, which involves predicting the likelihood of a new link forming between two nodes in the network. Graph convolutional neural networks (GCNNs) have recently emerged as a powerful approach for link prediction, leveraging the graph structure and node features to learn effective representations and predict links between nodes. This chapter provides an overview of recent advances in GCNNs for link prediction in social networks, including various GCNN architectures, feature engineering techniques, and evaluation metrics. It discusses the challenges and opportunities in applying GCNNs to social network analysis, such as dealing with sparsity and heterogeneity in the data and leveraging multi-modal and temporal information. Moreover, this also provides reviews of several applications of GCNNs for link prediction in social networks.