alita9/muse-sarcasm-explanation
MuSe: Multimodal Sarcasm Explanation (Reformatted) This repository provides a Hugging Face-compatible version of the MuSe (MORE) dataset. Modifications in this version To make the dataset easier to use with the datasets library, the following changes were made: Unified Schema: Merged separate OCR and Non-OCR files into a single test split. Metadata Flags: Added an is_ocr (boolean) column to distinguish between image types. Image Integration: Converted image… See the full description on the dataset page: https://huggingface.co/datasets/alita9/muse-sarcasm-explanation.
MuSe: Multimodal Sarcasm Explanation (Reformatted)
This repository provides a Hugging Face-compatible version of the MuSe (MORE) dataset.
Dataset Description
- Original Authors: LCS2-IIITD (Shweta Desai, et al.)
- Original Source: GitHub Repository
- Paper: Nice perfume. How long did you marinate in it? Multimodal Sarcasm Explanation (AAAI 2022)
Modifications in this version
To make the dataset easier to use with the datasets library, the following changes were made:
- Unified Schema: Merged separate OCR and Non-OCR files into a single
testsplit. - Metadata Flags: Added an
is_ocr(boolean) column to distinguish between image types. - Image Integration: Converted image paths into a native Hugging Face
Imagefeature for direct loading. - Labeling: Explicitly labeled all samples as "sarcastic" to match standard classification formats.
Citation
If you use this dataset in your research, please cite the original work:
@inproceedings{desai2022nice, title={Nice perfume. How long did you marinate in it? Multimodal Sarcasm Explanation}, author={Desai, Shweta and Pawar, Tanmay and Shah, Parth and Mittal, Akshat and Das, Amitava and Shah, Rajiv Ratn and others}, booktitle={Proceedings of the AAAI Conference on Artificial Intelligence}, year={2022} }
License
This dataset is distributed under the MIT License, as per the original source repository.
