Version control commits capture a small change in programs which were made by a programmer with an intention. For example if the programmer implements a web shop he could add the feature 'user login' and commit these changes. There are thousands of repositories on git-hub, thus a huge amount of more or less well structured data. Git-hub also gathers other data such as programming language, popularity of repository and much more. The programmer also adds a comment.
There have been research in genetic programming where a search in the space of possible programs is performed. This space is very high-dimensional and small changes in the code can have a huge impact on the program output, which makes it very difficult. However, version control histories available on git-hub could be used to guide and facility this search. Is there research in this direction?
[link][6 comments]