CAME (Convolutional Analysis of code Metrics Evolution) is a deep-learning based anti-patterns detection approach. CAME uses a Convolutional Neural Network architecture to detect anti-patterns from both structural and historical information mined from version control systems (e.g., Git).
This approach has been experimented for the detection of God Class.
If you want to train/test CAME on other java software systems or using a different set of metrics, you will need to produce some metric files in order to build the input matrices for our CNN. This can be done with the RepositoryMiner component.
This repository is organized as follows:
- approaches: The source code of the God Class detection approaches investigated in this work, i.e., CAME, MLP, decision tree, svm, DECOR, HIST and JDeodorant.
- data: Contains necessary data to run the experiments.
- antipatterns: The oracle, i.e., manually-tagged occurrences of God Class in eight software systems.
- metric_files: Files containing the metrics related to code components. Each file corresponds to one commit in the history of the system.
- data_construction: Code used to generate the metric files through the repository_miner component.
- experiments: The source code of our experiments: training, comparison, parameter tuning, etc.
- utils: Modules used to access and manipulate the data.
The paper associated to this repository has been accepted for inclusion in the Research Track of the 35th IEEE International Conference on Software Maintenance and Evolution (ICSME), 2019.