unibz-core/mining-cms-patterns

Mining-CMs-Patterns is an application for mining patterns in conceptual models encoded in different conceptual modeling language. Reference paper: Pattern Discovery in Conceptual Models Using Frequent Itemset Mining; Fumagalli, Mattia; Sales, Tiago Princes; Guizzardi, ER 2022

Mining-CMs-Patterns

Mining-CMs-Patterns is an application for mining patterns in conceptual models encoded in different conceptual modeling language.

The current version of the application is able to parse and mine any UML Class Diagram Model encoded in a {.json} format, which can be generated through the OntoUML plugin for Visual Paradigm (free community edition).

The main scope of Mining-CMs-Patterns is to support knowledge engineers in the empirical discovery of modeling patterns.

The application is written in Python, requires Python 3.6+, and has been tested on Mac OSX.

Installation

• First, start by cloning this repository.

git clone https://github.com/unibz-core/Mining-CMs-Patterns

• We recommend to use virtualenv for development.

• Once the virtual environment is created, install the python dependencies on the virtual environment.

pip install -r requirements.txt

• Access the application folder.

cd scripts

• Start the application.

python3 main.py

Get Started with Mining-CMs-Patterns

• Interact with the command line by providing the required inputs, e.g.:

Welcome to our pattern discovery application!
Press Enter to continue...
Please, select the weight of generalization relations,
('integer' from 0 to 9): 1
Please, select the weight of associations,
('integer' from 0 to 9): 3
Enter the reference number of nodes for generating graph partitions
(value ≥ 3 suggested): 5   

• Remember to put the models to be mined in the models folder.
• All the files we used in the experiment, currently collected in the models folder, where generated by pulling the OntoUML Repository files in the folder importing/ontouml-models-master/. All the files can be renamed and extracted in a single folder by running the fileimport.py script.
• The output of the mining process will be created in the output folder.
• If you want to restart the process remember to delete all the created files through the related command.

Want to clear folders? (Y/N): Yes!

Papers

Fumagalli, Mattia; Sales, Tiago Princes; Guizzardi, Giancarlo; Pattern Discovery in Conceptual Models Using Frequent Itemset Mining,  ER 2022, International conference on conceptual modeling