Cascading Impact Defense Data

该数据集包含了不同规模的图形测试案例（包含25、100和1000个节点的图形），旨在评估在对抗性图形环境中减轻级联效应的提议方法的性能。数据集不仅包含实际数据，还通过特定方法生成了反事实数据，以探索有效组合动作的子动作集。这样的反事实数据生成显著增加了数据集的大小，有助于验证所提方法的性能。该数据集覆盖了从小型到大型图形的规模（25、100、1000个节点），其任务是对在级联图形场景中的防御策略进行基准测试。

This dataset comprises graph test cases with varying scales (graphs containing 25, 100, and 1000 nodes respectively), and is intended to evaluate the performance of proposed methods for mitigating cascade effects in adversarial graph environments. The dataset includes not only real-world data but also counterfactual data generated via specific methods to explore sub-action sets of effective combined actions. Such counterfactual data generation significantly increases the dataset's size, which helps verify the performance of the proposed methods. Covering scales from small to large graphs (25, 100, and 1000 nodes), this dataset is designed to benchmark defense strategies in cascade graph scenarios.