On the Evaluation Metrics of Automated Program Repair

Recent research has made promising progress on automated program repair, but there still lacks common evaluation metrics in the evaluation process, increasing the risk of drawing inconsistent or even contradict conclusions. Based on the survey of recent repair work on automated repair published in five main venues, this paper conducts the discussion on the evaluation metrics having been introduced in literature, and in the furure we plan to present our suggest on how to construct the common metrics to facilitate the fair comparisons among competitive repair techniques.