Optimizing viral marketing strategies via a modified human evolutionary metaheuristic for social network influence

With the emergence of the social networks, the entire aspect of information distribution has been transformed and the dissemination of ideas, products and innovations has become fast. An extremely relevant issue in this undertaking pertains to the issue of maximization of influence with an aim to achieve a minimal set of seed nodes that may be used to maximize information spreading at the networks. The paper introduces a novel Discrete Modified Human Evolutionary Optimization (DMHEO) algorithm that has been optimized to perform viral marketing in large-scale social networks. DMHEO is based on effective methodology of initialization and pre-selection of leaders to enhance convergence and local optimum is avoided. The proposed method is confirmed to work on real-world network data: dimacs10-netscience and ego-Facebook, on Independent Cascade and Weighted Cascade models, to compare the results with the above mentioned algorithms. Comparison experiments indicate that the suggested DMHEO is more effective than other state of the art algorithms like Discrete Particle Swarm Optimization, Discrete Bat Algorithm and Discrete Bat Algorithm Modified in terms of influence spread and robustness. Additional conclusive studies on convergence, efficiency in calculations and reduction of errors will be done to determine its scale and effectiveness. The findings suggest that DMHEO would be a good optimization model to consider using by marketers who want to achieve maximum reach and interactions in the dynamic social network environments.