Research group
Machine Learning Methods in Software Engineering
Code Style Embeddings
Active
This project is aimed at designing a system that builds numerical representations of developers' individual coding styles in an unsupervised manner, based on data from version control.
Building embeddings of individual code style is one way to enable team collaboration tools to detect and track knowledge transfer processes in teams. While it looks as a far-fetched goal, this project is our take on this problem.
The paper is to appear in CHASE 2020 proceedings.
Participants
Publications
Authorship Attribution of Source Code: A Language-Agnostic Approach and Applicability in Software Engineering
August 2021
Egor Bogomolov, Vladimir Kovalenko, Alberto Bacchelli, and Timofey Bryksin
Building Implicit Vector Representations of Individual Coding Style
June 2020
Vladimir Kovalenko, Egor Bogomolov, Timofey Bryksin and Alberto Bacchelli