A Survey on Automated Program Repair Techniques

• DOI: 10.48550/arxiv.2303.18184
• Publication Date: 2023-01-01
• Authors: Kai Huang; Zhengzi Xu; Yang Su; Hongyu Sun; Xuejun Li; Yan Zheng; Yuqing Zhang
• Keywords: Popularity

With the rapid development and large-scale popularity of program software, modern society increasingly relies on software systems. However, problems exposed by have also come to fore. Software defect has become an important factor troubling developers. In this context, Automated Program Repair (APR) techniques emerged, aiming automatically fix reduce manual debugging work. particular, benefiting from advances in deep learning, numerous learning-based APR emerged recent years, which bring new opportunities for research. To give researchers a quick overview techniques' complete future opportunities, we revisit evolution discuss depth latest paper, is introduced terms four different patch generation schemes: search-based, constraint-based, template-based, learning-based. Moreover, propose uniform set criteria review compare each tool, summarize advantages disadvantages techniques, current state development. Furthermore, introduce research related technical areas that provided strong motivation advance Finally, analyze challenges directions, especially highlighting critical large language models