Empirical analysis on uncertainties in social networks towards information diffusion

Abstract

Social Networks contain a structure with linkage of nodes connected by means of their edges between them. The value of each node varies depending upon their nature. Accessing the significant nodes plays an essential role in the Information Diffusion process. Node with the higher tie strength or connections with them are said to have highest order of spreading information within the social networks. So, there is a need to find influential node with least cost and time to enhance Information Diffusion. Hence, an Information Spreader Enhancer system is proposed, which aims to identify influential nodes in communities, predicts links between the users within the Community to improve tie-strength between the nodes thereby increases diffusion and correlation between the influential nodes and other nodes for boosting precise information spread within the network. This is defined by considering centrality, weights of edges, nodal metrics and other properties of nodes such as local property, global property and location. Considering and improving these parameters will result in improvement of Information Diffusion process. The Information Spreader Enhancer system has three major parts Influential node detection, Link Prediction and Correlation based Information spreader. The first approach Support Vector Bayesian Machine (SVBM) mechanism is proposed for finding the influential node. In this work, a hybrid strategy is proposed which uses power dissipation of nodes and learning method - SVBM. The centrality metrics of the nodes namely closeness centrality and degree centrality is taken with distance metrics Euclidean and Hamming distance between them. These metrics are then combined with coefficient value of nodes obtained through Pearson correlation coefficient to give influential range of each node. newline newline newline