| --- |
| license: cc-by-4.0 |
| --- |
| # Morables-PD Dataset |
|
|
| ## Description |
|
|
| This repository contains a Public Domain, cc-by-4.0 release of the data used in "<span style="font-variant: small-caps;">Morables</span> : A Benchmark for Assessing Abstract Moral Reasoning in LLMs with Fables", presented at EMNLP 2025. |
|
|
| This parallel release only contains the title-fable-moral triples within Public Domain, with the annotator verified morals. |
|
|
| Each fable has an associated free-text moral, sourced from various websites and books (detailed in the paper). It is intended for use in NLP text understanding and moral inference tasks. |
|
|
| ## Contents |
| - **File Format**: JSON |
| - **Number of Records**: 308 |
| - **Content of Each Entry**: |
| - `alias`: (str) Unique ID of the fable |
| - `title`: (str) Title of the Fable |
| - `story`: (str) Content of the fable / short story |
| - `moral`: (str) Moral of the story |
| - **List of files**: |
| - `Morables-PD.json` : story/fable/moral triples, restricted to Open Domain translations; |
|
|
| ## Citing this dataset |
|
|
| If you use this dataset in your research, please cite the main reference paper: |
|
|
| ``` |
| @inproceedings{marcuzzo2025morables, |
| title={Morables: A Benchmark for Assessing Abstract Moral Reasoning in LLMs with Fables}, |
| author={Marcuzzo, Matteo and Zangari, Alessandro and Albarelli, Andrea and Camacho-Collados, Jose and Pilehvar, Mohammad Taher}, |
| year={2025}, |
| booktitle={Proceedings of EMNLP}, |
| publisher={Association of Computational Linguistics}, |
| } |
| ``` |
