READ-ME.txt

The study considered four software systems, written in two distinct programming languages (Visual Basic and Java). We selected a total of ten participants, for each of the two participating groups, we randomly selected one system in Java and one system in Visual Basic, so that in the end each group analyses systems in both languages. Then, each participant started the architecture recovery procedure by applying two reengineering approaches per session, which produced two models representing the modular decomposition of the system.

In the first part of the session, the participant applied only one technique using either a semi-automatic decomposition based on software visualization or an automatic decomposition based on software clustering.   After obtaining the first model, the other approach was presented and it was requested to produce a new model, based on the two techniques.  We also made sure that the application order of each technique was prioritized in one of the two systems in each group.  In this way, it was possible to calculate the accuracy of the resulting models using one technique, two techniques while varying their order of application.
 
The resulting models are available at this repository. Also, in the folder "ScreenShots_visualizationTools" there is examples of the models obtained from software visualization tools used by the subjects. In the folder "BloxPlots" there is bloxplots obtained from the results of experiments.