Predicting Information Cascade on Twitter Using Random Walk

Abstract

Abstract Online Social networking (OSN) platforms are the most widely used platforms for spreading information about current topics. One of the major advantages of these websites is the accessibility and reach to the users of these platforms. These platforms do not demarcate information cascading based on geography, race, caste, and creed to name a few. The prediction of a cascade is an important problem as it gives a deep understanding of opinion-shaping in audience. For eg. diffusion about a product, politics for, example, celebrity popularity. Many models have been proposed for solving this problem that uses content and user-based information, temporal and structural features of the network. Propagation is faster if the similarity between entities is high. In this paper, we propose a random walk based method that also exploits similarity measures (user and content) to predict the diffusion cascade. We evaluated our method using Twitter as a use case and demonstrated that our method outperforms the existing structural and content-based methods being proposed earlier.