Competitive Influence Maximization on Online Social Networks: A Deterministic Modeling Approach

The Competitive Influence Maximization (CIM) problem has been extensively studied in recent years due to its importance in Viral Marketing. Basically, CIM problem requires finding out a set of seed users to diffuse information such that the number of affected nodes could be maximized in a competitive context where other competitors are also performing information broadcasting activities. Recent works focus on uncertainty models (probabilistic models) to address this problem at the cost of difficulty which is #P-hard in estimating objective functions, thus those algorithms are time consuming even in the case of small or medium size networks. In reality, with the support of data mining techniques, we are able to build deterministic information spreading models. In this work, we propose a new approach to the CIM problem based on such a deterministic information spreading model. Our approach shows that the complexity in estimating objective functions is O(n²) and it also proposes 2 effective algorithms to address CIM problem. Experiment results on 4 benchmark datasets Email-EU, Gnutella-1, Gnutella-2, and Epinion show that our algorithms outperform other algorithms, especially in term of running time on medium size networks.