CoIM: Community-Based Influence Maximization in Social Networks

Abstract

Influence maximization (IM) is the problem of identifying k most influential users (seed) in social networks to maximize influence spread. Despite some recent development achieved by the state-of-the-art greedy IM techniques, these works are not time-efficient for large-scale networks. To solve time-efficiency issue, we propose Community-based Influence Maximization (CoIM) algorithm. CoIM first partitions the network into sub-networks. Then it selects influential users from sub-networks based on their local influence. The experimental results on both synthetic and real datasets show that proposed algorithm performs better than greedy regarding time with the almost same level of memory-consumption and influence spread.