COMODI (COunting MOdel DIfferences) is a tool for model comparison. It provides several algorithms and distance metrics to compare 2 or more models (dot or xmi files). Given a folder of models, COMODI creates distance matrices comparing all the models inside that folder. Then, a hierarchical clustering is performed and the most representative models of the folder are automatically chosen.
A human readable graphical representation is also plotted. It easily shows the coverage of solutions’ space of a model set. Finally, some extra statistics are computed over the input folder (closest models, average distance).
Get it and use it
COMODI is free and it usage is simple. Download it (jar, all scripts, examples)
Get Comodi here. This archive contains:
- COMODI.jar to compare two models (xmi or dot files).
- COMODI.sh to handle a folder of models.
- Matrix clustering script.
- R scripts (clustering, voronoi).
You can find also many examples here
Linux (or mac are prefered), R software (+ tripack, tikzdevice packages)