Automating code review activities by large-scale pre-training

Code review Code smell KPI-driven code analysis Code (set theory) Benchmark (surveying) Software inspection
DOI: 10.1145/3540250.3549081 Publication Date: 2022-11-09T20:46:22Z
ABSTRACT
Code review is an essential part to software development lifecycle since it aims at guaranteeing the quality of codes. Modern code activities necessitate developers viewing, understanding and even running programs assess logic, functionality, latency, style other factors. It turns out that have spend far too much time reviewing their peers. Accordingly, in significant demand automate process. In this research, we focus on utilizing pre-training techniques for tasks scenario. We collect a large-scale dataset real-world changes reviews from open-source projects nine most popular programming languages. To better understand diffs reviews, propose CodeReviewer, pre-trained model utilizes four tailored specifically evaluate our model, three key related activities, including change estimation, comment generation refinement. Furthermore, establish high-quality benchmark based collected data these conduct comprehensive experiments it. The experimental results demonstrate outperforms previous state-of-the-art approaches all tasks. Further analysis show proposed multilingual benefit reviews.
SUPPLEMENTAL MATERIAL
Coming soon ....
REFERENCES (48)
CITATIONS (90)
EXTERNAL LINKS
PlumX Metrics
RECOMMENDATIONS
FAIR ASSESSMENT
Coming soon ....
JUPYTER LAB
Coming soon ....