Price Competition of Spreaders in Profit-Maximizing Sponsored Viral Marketing

Abstract

In online social networks, celebrities are usually paid to promote products via posting or forwarding ads or related information. Imagine that one day when everyone is allowed to register as a spreader and participate in the campaign to sell influence, how much money should be claimed? Two factors play vital roles in deciding the price. One is how influence is valued by buyers (advertisers). The other is how one's price is affected by that of others. In this paper, we consider that the influence is valued as the number of final 'activations' under some existing information diffusion processes, and focus on the latter, namely, the price competition. We model the scenario as a pricing game where spreaders compete with each other under selection policies of the advertiser, who is trying to maximize its profit. We draw conclusions for three cases of the advertiser. First, an omniscient advertiser always selects the optimal set of spreaders. We show that the competition is so fierce that each spreader can only claim its unique influence in the Nash equilibrium (NE), and the equilibrium is also unique. Second, the greedy advertiser selects spreaders using the simple greedy algorithm. We deduce that the unique NE exists when the number of spreaders is less than four; however, the existence of NE cannot be guaranteed when there are at least four spreaders. Third, the advertiser adopts a 'double-greedy' method that greedily selects spreaders one by one in accordance with their registration order. We conclude that the unique NE exists and the utility of the platform is at least 1/2 to the optimal and also bounded by 1/2 to the influence of all spreaders. © 2014 IEEE.