Balancing Centrality and Similarity for Efficient Information Recommendation in Social Networks

To maximize the scope and effect of information propagation in social networks, we need to explore both its intrinsic properties and spread theories elaborately. The key challenge is how we find the individuals who are the most powerful to maximize the propagation of a specific thing in a given network topology. Due to its high time complexity, the classical solution, i.e., Greedy algorithm, can not be applied to solve this problem, especially for a large-scale social network. In this paper, taking centrality and similarity into account, we define a novel viral marketing model to update the opinions of individuals on specific things or products, and propose an advertisement recommendation scheme to find the optimal individuals for influence maximization based on two epidemiology models, i.e., SIR (Susceptible, Infectious and Removed) and IC (Independent Cascade). Through extensive simulations and analysis, we show that the proposed algorithm can improve the performance of the recommendation system with a low time complexity, and the running time of our proposed algorithm is around 40% lower than that of the benchmark.