EVENTO
Understanding and Managing the Diffusion of Information in Online Social Networks
Tipo de evento: Seminário LNCC
Online social networks provide unprecedented amounts of information about social interactions and therefore enable the study of various social problems on a scale and at a level of detail that has never been possible before. In this talk, we will consider a central challenge in this problem space: understanding and managing the diffusion of information in online social networks. We will present two different applications and approaches to demonstrate the complexity of this problem space and to pinpoint the methodologies that will navigate us through it. First, we will consider the problem of limiting the spread of misinformation and convey the power of theoretical approaches for tackling this challenging problem. To this end, we will introduce models of diffusion that capture the notion of competing campaigns and develop protocols that limit the spread of misinformation under such models. Our solution relies on identifying a subset of individuals to start spreading the good information campaign from. We show that while identifying the optimal set of such individuals is NP-hard, an approximation can be provided in polynomial time. Next, we will describe a data-driven approach for modeling adoption behavior in online social networks. By employing the power of big data and statistical methods, we challenge the common use of local influence models and revive theories introduced in diffusion of innovations research to model human behavior. Our work moves beyond the conventional wisdom that models humans as mere beings who are only aware of their immediate surroundings and as a result provides a significant improvement for various prediction tasks. The value of this technique is demonstrated through a cross-network, cross-metric, cross-model evaluation framework. The two studies exemplify the complexity and diversity of challenges faced in complex systems such as online social networks.
Data Início: 20/05/2013 Hora: 14:00 Data Fim: Hora: 15:00
Local: LNCC - Laboratório Nacional de Computação Ciêntifica - Auditorio A
Comitê Organizador: Ceren Budak - University of California Santa Barbara - UCSB -
