Hazardous Influential Node Identification in Online Social Networks

Abstract

Online Social Networks (OSN) play an inevitable role in many real-time applications. Digital marketing uses the influential node to share the information. The influential node is more hazardous at times in many aspects. The proposed Hazardous INFluential (HINF) algorithm works by identifying the most influential hazardous node in the OSN. The HINF is executed based on the propagation models. The HINF aims to identify the hazardous influential node and block the activity. Also, it attempts to automate the process. Firstly, the Random Walk (RW) cluster, the OSN, and the TRusted Node (TRN) are identified for each cluster. Secondly, the influential node is identified using the oscillation score and the Information Sharing Rate (ISR) considering the network delay and resistance. Thirdly, the rank score finds the top-k influential node in OSN. Task-1 identifies the most influential node. Task-2 finds the hazardous node among the most influential nodes. Fourthly, the hazardous nodes are identified using the customized PyTorch classifier. Finally, it is cross-checked using the calculated cosine and Jaccard similarity score and the activity is blocked. The accuracy and F1-Score are analyzed for performance evaluation on the extracted Facebook dataset. The results prove that the proposed HINF outperforms the existing works.