Link injection for boosting information spread in social networks

Social media have become popular platforms for spreading information. Several applications, such as ‘viral marketing’, pause the requirement for attaining large-scale information spread in the form of word-of-mouth that reaches a large number of users. In this paper, we propose a novel method that predicts new social links that can be inserted among existing users of a social network, aiming directly at boosting information spread and increasing its reach. We refer to this task as ‘link injection’, because unlike most existing people-recommendation methods, it focuses directly on information spread. A set of candidate links for injection is first predicted in a collaborative-filtering fashion, which generates personalized candidate connections. We select among the candidate links a constrained number that will be finally injected based on a novel application of a score that measures the importance of nodes in a social graph, following the strategy of injecting links adjacent to the most important nodes. The proposed method is suitable for real-world applications, because the injected links manage to substantially increase the reach of information spread by controlling at the same time the number of injected links not to affect the user experience. We evaluate the performance of our proposed methodology by examining several real data sets from social networks under several distinct factors. The experimentation demonstrates the effectiveness of our proposed method, which increases the spread by more than a twofold factor by injecting as few as half of the existing number of links.