Yifan Wang

The paper introduces Repair-Augmented Constraint Learning (RACL), a framework that models contextual decisions by allowing systems to learn whether a candidate should be repaired before being vetoed, significantly reducing false vetoes compared to existing methods.