Marketing and privacy under information diffusion in online social networks

Abstract

The online social network provides a platform for people to share information and experience with their friends and followers. People's sharing behavior may help them achieve benefits from the social network, and help advertisers spread advertisements through social links, but may also bring about privacy problems. How to monetize the online social network for the advertiser and how to help users protect privacy are two main topics of online social network research. This thesis develops the frameworks to model these problems and provides strategies for viral marketing and privacy protection.

Firstly, the trade-off between utility and privacy cost is studied. When asking questions in an online social network, the potential answerers will give people utility. However, people's personal and other sensitive information may be exposed when the question spreads via the social network, which will result in privacy cost. People will always incur privacy cost when trying to obtain utility. To solve this problem, a framework is proposed to quantitatively evaluate the utility and privacy cost in online social search, and an algorithm is designed for users to gain more utility under the same privacy cost.

The second part of the thesis investigates how to design pricing strategies for advertisers to gain more revenue through online social network marketing. Different from the existing study of pricing schemes, we study how the price may influence the diffusion. A framework which incorporates the influence maximization problem and a multi-state diffusion model is proposed. In the diffusion model, users are divided into different groups by their purchasing behavior and have different influence power, which informs how pricing strategies can influence the potential buyers. Different pricing strategies are designed with different pricing sequence orders and different promotion times, and are compared by using the framework. Some guidelines for the sellers are proposed based on the simulation results of different strategies.

Finally, the thesis addresses the incentive strategy in online social network marketing. Considering the incentive cost, the revenue function is no longer monotone, and most methods for influence maximization problem cannot be used. An algorithm to find seed nodes is designed for the revenue maximization problem when considering the incentive strategy. Moreover, the thesis also explores how the incentive strategy can influence the diffusion process. The incentive strategy is evaluated by the framework with a multi-state diffusion model and can be used to explain how people's purchasing behavior is influenced. The simulation shows that the incentive marketing strategy performs well in activating more people and achieving more revenue.