Cost-effective competition on social networks: A multi-objective optimization perspective

Abstract

The competition on social networks has attracted widespread attention in recent years. Despite the great progress in this domain, acquiring more votes for competitors with fewer recruiting costs is still a research blank. This paper addresses this issue from three aspects: modeling, proposition, and methodology. First, a succinct competitive model is built to clear the final beliefs of normal agents. Then, a cost-effective competition (CEC) problem is proposed by formulating the number of votes and the recruiting costs as two optimization objectives. Seeing the cost constraints, a theoretical cost budget is also determined. Next, a multi-objective optimization approach (MOCEC) with theoretical guarantees is developed to solve the CEC problem. In the experiments, MOCEC is compared with seven prevalent competitive methods on ten real-world social networks. We show that MOCEC can achieve the best trade-offs between maximizing the number of votes with minimizing the recruiting costs in all test environments. Especially when the number of votes reaches the maximum value, MOCEC can save up to 50% of the recruiting costs than the best baseline.