Latent Viral Marketing, Concepts and Control Methods (1704.01775v1)

Abstract: Numerus works that study the spread of information in social networks include a spreading mechanism in which a set of nodes is initially infected (i.e. seeded), followed by a viral process, which spontaneously spread the message through the nodes of the network. These models are used to describe the spread of rumors as well as the spread of new products and services. In reality however, it is quite rare that a product or service spreads through a social networks solely by viral forces. It is more common, that the spreader would invests a continuous active effort of its sales representatives in order to enhance the spread. The Latent Viral Marketing Model is a spreading model that fits this reality. Along the description of the model, the paper continues and proposes a simple Scheduling of Seeding heuristics, which recommends the node to seed at each period. A large study of empirical simulations shows that under a wide range of realistic initial conditions, the Scheduling Seeding method improves a product adoption rate by 29%-94% in comparison to existing state-of-the-art seeding methods.