For the assignment, you will reverse engineer a Java system of your choice. Your first task will be to select a system to work with. The only requirement is that the system be large enough so that clustering it makes sense (this will depend on your choice of what the entities to be clustered are). You can look for open source systems of interest in the following sites:

1. sourceforge
2. freshmeat
3. koders
4. Google code search

Once you have selected a system, make sure you are able to build it locally. Then, extract dependency information from it using either the Javex extractor. You may need to write some Grok scripts to create a fact base that contains exactly the facts you will need.

Next, cluster the set of entities of the system with 10 different algorithms:

• ACDC
• 3 different configurations of Bunch (you may choose which ones). Do not use the exhaustive algorithm as it is too slow. The genetic algorithm may also take a long time to finish.
• 6 different configurations of hierarchical clustering algorithms (you may choose which ones)

Next, compute all 90 values of MoJo distance and KE recall rate between all pairs of decompositions created above.

Finally, select one of the decompositions created above, and create a landscape of the system. Re-arrange the landscape in order to create a view that would be helpful to someone attempting to understand this system.

For this assignment, you will work with a variety of tools.