Optimizing Alignment with Less: Leveraging Data Augmentation for Personalized Evaluation