Context-Aware AI-Powered Code Review Assistant

Software developers adopt modern code review, a software quality assurance activity, to ensure that software has high quality. Recent work has proposed various AI-powered code review assistants. However, these approaches still lack contextual information to generate an explanation of their predictions and recommend the improved version of the submitted patch. Thus, we hypothesize that AI-powered code review assistants should not solely rely on technical information, but also the contextual information of its technical information. To validate this hypothesis, we design, propose, develop, and evaluate context-aware AI-powered code review assistants. The experimental results confirm our hypothesis.