A Combined Approach for k-Seed Selection Using Modified Independent Cascade Model

Abstract

Seed set selection for Information Maximization Problem (IMP) is an interesting NP-hard problem. It plays an important role in understanding the spread of information. K-shell decomposition is a graph degeneracy approach to find core as a seed set. But if the cascade of information is controlled by a hop limit and the number of nonoverlapping neighbors between the nodes is less, then the information is unable to spread quickly in the network. Here, our objective is to investigate the fast influence maximization. Some of the information is of no use after some time, in such cases, limited hop diffusion is a better model of information diffusion. Also in most of the diffusion model, the probability of diffusion is kept as a constant. But in the real-world scenario, this may be different. Hence, we have proposed a Modified Independent Cascade (MIC) model that diffuses information with different probabilities of influences of the shells and with a limiting propagation. The proposed method selects the nodes of highest degree from each shell. The proposed approach outperforms the random seed set selection in all cases. But it outperforms the k-core in case of minimum nonoverlapping neighbors. The proposed method is investigated upon the MIC model.