Citation Key: wang2021b <标题>: 通过评分量表进行简答题评分的数据增强 <AI Smry>: This paper proposes a semi-supervised method to train a neural SAG model that extracts keyphrases that are highly related to answers scores from rubrics, and demonstrates that both performance of grading and justification identi-cation is improved by integrating attention semi- supervised training, especially in a low-resource setting.