Popis tématu

Oborový projekt v oboru Softwarové inženýrství.

Automatická klasifikace commitů

Ve spolupráci s kolegy z OTH Regensburg hledáme zájemce s alespoň základními znalostmi klasifikačních metod postupy strojového učení, pro řešení dále popsaného úkolu. Další informace u mne nebo u Petra Píchy (UN308).

Version Control Systems (VCS) play a central role in software
engineering processes, as they track changes in software. Many of them,
like the most popular one, git, store besides annotations of textual
changes of the code (the diff or patch) an additional message that
informally describes the changes in a human readable manner (the commit
message).

Generally, there is no information on what type of change (e.g.,
'Feature', 'Bugfix', 'Refactoring', 'Cleanup', ...) a commit introduces.
Being able to classify the type of a commit allows for further analyses,
such as determining the consistency of a project, filtering for certain
types of commits or even to draw conclusions on the maintenance or
stability of the project.

In this project, we are interested in automatically classifying commits
into categories with the use of state of the art machine learning
techniques. The task is to evaluate different machine learning
algorithms, to balance their pros and cons, to determine a sound feature
set as well as the accuracy of the chosen method. A proof of concept
implementation is already existing.

Téma vypsal: Doc. Ing. Přemysl Brada, MSc., Ph.D. (UC 354)

Vypsáno pro akademický rok 2018/2019 dne: 2018-07-06

Zadáno komu: (Smazané)

Aktualizováno 0