Random sampling baseline: randomly removing edges to match our framework's reduction percentage provides a critical control for evaluating whether intelligent filtering outperforms naive approaches.