Reconstructing Diffusion Model for Virality Detection in News Spread Networks

In today's competitive world, organizations take advantage of widely-available data to promote their products and increase their revenue. This is achieved by identifying the reader's preference for news genre and patterns in news spread network. Spreading news over the internet seems to be a continuous process which eventually triggers the evolution of temporal networks. This temporal network comprises of nodes and edges, where node corresponds to published articles and similar articles are connected via edges. The main focus of this article is to reconstruct a susceptible-infected (SI) diffusion model to discover the spreading pattern of news articles for virality detection. For experimental analysis, a dataset of news articles from four domains (business, technology, entertainment, and health) is considered and the articles' rate of diffusion is inferred and compared. This will help to build a recommendation system, i.e. recommending a particular domain for advertisement and marketing. Hence, it will assist to build strategies for effective product endorsement for sustainable profitability.