A Fast Algorithm for Competitive Recommendation Marketing Strategy

The development of social networks changes the way of information diffusion and the means of selling and buying. In this paper we study the problem of online promoting for a product with limited budget in the competitive propagation social networks, we called this problem competitive recommendation marketing strategy (CRM). The key of CRM problem is to find out a fixed number of target nodes as the original recommended nodes (by giving them free or discount samples) of our product, that can maximize the diffusion range and number of adoptions in the competitive propagation models. We first introduce two kind of competitive information propagation models called Competitive Independent Cascade model (CIC model) and Competitive Linear Threshold model (CLT model). Then we demonstrate the submodular property of CRM problem in CIC and CLT moels. Finally, we give greedy and fast heuristic algorithms for this problem. Experiment results show the efficiency of our algorithm.