Learning-Anti-Patterns

data and scripts used for the experiment we executed in the Onto.Com 2021 paper

Quickstart Guide

Install python 3.x.

Install the following libraries:

$ pip install networkx
$ pip install matplotlib.pyplot
$ pip install regex
$ pip install pandas
$ pip install pygraphviz

Install RapidMiner free version.

Run the the gufo2alloy converter to generate Alloy .als specifications (see related model examples).

Customize the simulations scope. Run simulations. Save the output simulations into .DOT files.

Copy any generated anti-pattern occurrence .DOT into a .txt file (see example file car-binover-proc.txt). Then execute the following steps:

1. Use the anti-pattern occurrence (.txt format) as input of generalize-pattern.py in scritpts

2. Use all the outputs of 1. to generate an example set like examples.xlsx.

3. Give examples.xlsx as input of RapidMiner embedding+clustering.xml process in scritpts - this is necessary to embed, bootstrap, evaluate, cluster and extract the clusters prototypes as described in the paper.

4. To generate the generalized anti-pattern give the output of 1. to generate-dot.py and then to visualize-graph.py.