Towards Automated Support for Eective Modern Code Review Activities

This thesis focuses on improving the efficiency of Modern Code Review (MCR), a widely adopted quality assurance practice in software development. MCR can be labor-intensive, especially as projects grow in complexity, leading to challenges in balancing reviews with development tasks. To address this, the thesis introduces several automated solutions: REVSPOT, which uses machine learning to identify lines likely to receive comments or require revision; COCHANGEFINDER, a GNN-based tool that recommends co-changed functions; and COMMENTFINDER, which suggests review comments. The thesis also presents a study on the impact of MCR on build outcomes at Atlassian, highlighting automation’s role in enhancing review efficiency.