IN ONLINE SOCIAL NETWORK USING GREY WOLF AND DEEP LEARNING TECHNIQUE FOR INFLUENTIAL USER PREDICTION (IUP)

Abstract

The far-reaching use of Online Social Networks (OSNs) and the often growing volume of knowledge provided by their members have motivated both corporate and scientific researchers to investigate how certain systems can be manipulated. According to recent findings, monitoring and evaluating the influence of OSN users has significant applications in the fields of health, economics, education, politics, entertainment, and other fields. The propagation model has an impact on a centrality measure's capacity to show a node's ability to disseminate influence. In certain modeling techniques, the centrality measures perform well on directed contacts. However, centrality measures not perform well for indirect contacts. To improve prediction performance, additional measures and combined centrality measures are proposed by employing linear combinations of measures is proposed in this article. The deep learning-based CNN algorithm is developed for Influential User Prediction (IUP). The relevant measures selected by Grey Wolf Optimization (GWO) are fed into Convolutional Neural Network (CNN) for training and trained model for IUP. GW algorithm initializes many grey wolves, finds the optimal measures and provides the best solutions (IUP) using CNN. The GW positions are updated to look for new solutions until a near-optimal solution is found. The proposed GW-CNN is compared with CNN, CPPNP, and TDSIP and proved GW-CNN provides the best results for IUP.