Short-term positive influence maximization problem in signed networks

Abstract Influence maximization problem is a classic problem in complex networks analysis. While most studies focus on unsigned networks with only positive relationships, real-world networks often involve both positive and negative relationships, known as signed networks. Additionally, the time factor is crucial in situations requiring rapid information spread. This paper addresses the short-term positive influence maximization problem in signed networks, incorporating edge polarity and time constraints. Our findings reveal significant differences between short-term and long-term propagation: nodes that maximize long-term positive influence cannot perform well in a short time in signed networks. Furthermore, we propose a local evaluation metric of nodes—2-order positive degree. Experiments on Slashdot, Epinions, and Wikipedia demonstrate that the 2-order positive degree metric outperforms the compared centrality metrics in identifying short-term influential nodes, offering new insights for influence maximization in signed networks.