Code for the paper "Exploring Large Language Models Capabilities to Explain Decision Trees".
Presented at Hybrid Human Artificial Intelligence (HHAI) 2024.
A summary containing prompts and responses for all experiments can be found in the experiments file.
- Python >= 3.8
- Scikit-learn
- OpenAI API
- Help
python src/main.py --help
python src/main.py -h
- Show help message.
- Run experiments by order number
python src/main.py --experiment-number <experiment_number>
python src/main.py -n <experiment_number>
- Run the experiment with the given number.
- Valid
<experiment_number>
options: 1, 2, 3, 4, 5, 6, and 7.
- Run prompt experiments by number
python src/main.py --prompt-number <prompt_experiment_number>
python src/main.py -p <prompt_experiment_number>
- Run the prompt experiment with the given number.
- Valid
<prompt_experiment_number>
options: 1, 2, and 3.
- Run text experiments by number
python src/main.py --text-number <text_experiment_number>
python src/main.py -t <text_experiment_number>
- Run the text experiment with the given number.
- Valid
<text_experiment_number>
options: 1, 2, 3, and 4
If you use this code for your research, please consider citing this work:
@inproceedings{serafim2024exploring,
title = {Exploring Large Language Models Capabilities to Explain Decision Trees},
author = {Serafim, Paulo Bruno and Crescenzi, Pierluigi and Gezici, Gizem and Cappuccio, Eleonora and Rinzivillo, Salvatore and Giannotti, Fosca},
booktitle = {Proceedings of the Third International Conference on Hybrid Human-Artificial Intelligence},
year = {2024},
volume = {386},
pages = {64--72},
doi = {10.3233/FAIA240183}
}