5.1.1.4. Co-change analysis.
In order to identify co-changes, we
applied association rule mining techniques.
To do so, we firstidentified the file which, should it be changed, would cause a
change to the web API.
While VirtualBox makes use of WSDL for
its web services, the WSDL file is generated and not initially available under versioning. For this reason, we perform our analysis
on the file which is used to generate the WSDL interface, that is
“src/VBox/Main/idl/VirtualBox.xidl”.
We then grab all the commits
where this file is changed and treat each commit as a “transaction”
for the Apriori algorithm.
By doing so, we obtain a list of which
files are associated with (and thus, tend to co-evolve) which other
source files