Few-Shot Learning for Issue Report Classification

This repository contains notebooks for training and testing the classifiers for our participation in the tool competition organized in the scope of the 2st International Workshop on Natural Language-based Software Engineering.

Our fine-tuned SETFIT model is available on HuggingFace.

How to use

• NLBSE23_1_Preprocessing.ipynb: Runs preprocessing and saves output in order to be used from the other notebooks.
• NLBSE23_2_RoBERTa.ipynb: Trains and tests RoBERTa model in different settings
• NLBSE23_3_SETFIT.ipynb: Trains and tests SETFIT model in different settings

It is essential to run the notebook NLBSE23_1_Preprocessing.ipynb and store the outputs, in order to be able to run the others, as it creates and preprocess the datasets.

It is recommended to run the RoBERTa and SETFIT notebooks with a GPU. For any problem in reproducing the experiments, feel free to open an issue or contact the authors.

How to cite

To cite the manually annotated dataset:

@dataset{colavito_dataset_2023,
	title        = {Few-Shot Learning for Issue Report Classification},
	author       = {Colavito, Giuseppe and Lanubile, Filippo and Novielli, Nicole},
	year         = 2023,
	publisher    = {Zenodo},
	doi          = {10.5281/zenodo.7628150},
	url          = {https://doi.org/10.5281/zenodo.7628150}
}

To cite this repository

@software{colavito_code_2023,
	title        = {{Few-Shot Learning for Issue Report Classification}},
	author       = {Colavito, Giuseppe and Lanubile, Filippo and Novielli, Nicole},
	year         = 2023,
	url          = {https://github.com/collab-uniba/Issue-Report-Classification-NLBSE2023},
	version      = {1.0.0}
}

To cite the paper:

@inproceedings{Colavito-2023,
	title        = {Few-Shot Learning for Issue Report Classification},
	author       = {Colavito, Giuseppe and Lanubile, Filippo and Novielli, Nicole},
	year         = 2023,
	booktitle    = {2nd International Workshop on Natural Language-Based Software Engineering (NLBSE)}
}