The identification of a seed set to maximize information spread in a network is crucial, a concept known as Influence Maximization (IM). Elegant IM algorithms could naturally extend to cases where each node is equipped with specific weight, referred to as individual effect, to measure the node's importance. Prevailing literature has typically assumed that the individual effect remains constant during the cascade process. However, this assumption is not always feasible, as the individual effect of each node is primarily evaluated by the difference between the outputs in the activated and non-activated states, with one of these states always being unobservable after propagation. Moreover, the individual effect is sensitive to the environmental information provided by surrounding nodes. To address these challenges, we extend the consideration of IM to a broader scenario involving general networks with dynamic node individual effects, leveraging causality techniques. In our paper, we address this through the development of a Causal Influence Maximization (CauIM) algorithm. Theoretically, for CauIM, we present the generalized lower bound of influence spread and provide robustness analysis. Empirically, in synthetic and real-world experiments, we demonstrate the effectiveness and robustness of CauIM, along with a novel acceleration technique.

翻译：在社交网络中，识别种子集以最大化信息传播至关重要，这一概念被称为影响力最大化（Influence Maximization, IM）。优雅的IM算法可自然扩展至每个节点具有特定权重（即个体效应）以衡量节点重要性的场景。现有文献通常假设个体效应在级联过程中保持不变。然而，这一假设并非总是成立，因为每个节点的个体效应主要通过激活与非激活状态下输出结果的差异来评估，而传播后这两种状态中始终有一种不可观测。此外，个体效应对周围节点提供的环境信息高度敏感。针对这些挑战，我们借助因果推断技术，将IM的考量扩展至具有动态节点个体效应的一般网络这一更广泛场景。在本文中，我们通过开发因果影响力最大化（Causal Influence Maximization, CauIM）算法来解决这一问题。理论上，我们给出了CauIM算法影响力传播的广义下界，并进行了鲁棒性分析。在合成与真实世界的实验中，我们证明了CauIM算法的有效性和鲁棒性，并提出了一种新颖的加速技术。