Modeling the Spread of Influence in Social Networks

Social networks play an important role in connecting people and facilitating human social interaction. Influence can spread through a social network. “Word-of-mouth” and “viral marketing” effects have been widely exploited to promote new products and technological innovations. For example, when an individual adopts a new product and finds it useful, she or he would recommend it to her or his friends and colleagues. One of this individual’s friends takes the advice, and may also feel excited about the product and spread the words about it to her or his own friends. In such a way, social influence can help diffuse new products or ideas. Modeling the spread of influence in social networks is valuable to computer science because of its relevance to digital networks (e.g., online social networks), but this research also holds importance in other fields, such as epidemiology, physics, and social sciences. The goal of this work is to derive a mathematical model that can accurately predict the influence of individuals in social networks. Such an accurate model can help computer scientists in designing new network protocols, structures, and policies to facilitate the spread of influence or information. Previous works on modeling the spread of influence assume the status of a single node in the network is independent of the status of other nodes in the network. In social networks, this is not true. For example, two friends in a social network tend to both either favor a product or reject it, meaning that the two individuals are spatially related to each other. In our work, we investigate the spread of influence in social networks using a Markov model, which assumes that neighboring nodes are spatially dependent on each other. To simulate the spread of influence in a social network, we use several generated network topologies and a real co-authorship network of scientists working on network theory, where the relationship between nodes is that author X wrote a joint work with author Y. We develop simulation tools in C++ to replicate influence diffusion through these networks. Our results show that our proposed Markov model predicts the spread of influence in a social network better than previously proposed models. To the best of our knowledge, this is the first attempt in studying the spatial dependence among nodes to describe influence dissemination.http://opus.ipfw.edu/stu_symp2017/1042/thumbnail.jp