Neural Network for Link Prediction in Social Network

The Internet has greatly advanced research in social networks, leading to the emergence of link prediction as a significant area of study in social network analysis (SNA). Link prediction involves predicting the formation of new connections between nodes that do not currently exist. This research trend has enabled people to connect with others based on shared characteristics, leading to improved communication. To enhance the accuracy of link prediction, researchers have employed various machine-learning techniques, including supervised and unsupervised learning. Also, deep learning models were widely adopted. In this research work, we attempt to develop a new link prediction approach using an Artificial Neural Network. Firstly, we adopt the node similarity measures to assign scores to pairs of nodes based on available dataset features. Secondly, based on a neural network, we can predict the future links with the goal of improving the accuracy of link prediction. The experimental results based on Twitch dataset show that our approach has a high prediction accuracy.