Influence maximization in social networks with user attitude modification

The aim of influence maximization problem is to find a k-size seed set that has the maximum influence. In previous works the modification of user´s attitude is seldom paid attention to. However from the psychology research, we know that people´s opinions are affected by their friends. Base on this, we present a new Linear Threshold model with Instant Opinions (LT-IO). We devise an attitude function A^t_u that describes node u´s attitude at time t, and the broadcast attitude which is the attitude when a node becomes active. To simulate information propagation in real world, we define a trust threshold η to justify whether a node follows or opposes the influence from its neighbor. We propose a heuristic algorithm IMLT-IOA to solve our problem, prove its submodularity and monotonicity and then obtain its approximation ratio which is (1 - 1/e). To the best of our knowledge, this is the first work that focuses on the influence maximization with user´s attitude modification. To verify our IMLT-IOA algorithm, we conduct extensive experiments on a large data collection obtained from real social networks, the results show that IMLT-IOA reduces the running time and meanwhile keeps effectiveness comparing to other algorithms.