Institutional Repository of Vilnius University
Daugiaagentė sistema, naudojanti didelės kalbos modelius kodo peržiūrai automatizuoti /
Abstract
dc:descriptionCode review is a common practice in the software development process aimed at ensuring the quality of the codebase. However, due to the accelerating pace of software development, there is a growing need to improve the efficiency of this practice. Multi-agent systems based on large language models (LLM-MAS) are emerging as a promising tool for process automation tasks, and their effectiveness in code review is therefore being observed. This study aims to contribute to the improvement of multi-agent systems used in code review automation task by pursuing the objective of identifying suitable enhancement methods for a multi-agent system and evaluating the impact of partial automation, based on a multi-agent system, on code review practice. To achieve this, a literature analysis was conducted, identifying topology modification and implementation of peer review loops as the enhancement methods, which during experimental research were found to be effective for improving the system. Improved system with these methods was then integrated into the "Azure DevOps" code management system, which allowed to conduct expert testing. After which data was collected on the effectiveness of the system's application in a practical environment. The survey results not only demonstrated the system's positive contribution to code review but also helped to identify the direction and significance of further research for its practical implementation.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2025
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Minajevas, Paulius,
Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- lit
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD239031833&prefLang=en_US