Implementation for the paper "Does Commonsense help in detecting Sarcasm?", published at Insights from Negative Results in NLP workshop, EMNLP 2021.
Authors: Somnath Basu Roy Chowdhury and Snigdha Chaturvedi
# create a clean conda env
conda create -n commonsense-sarcasm python==3.8
source activate commonsense-sarcasm
pip install -r requirements.txt
Install COMET (trained on ATOMIC) package using the instructions provided in the following repository - https://github.com/vered1986/comet-commonsense.
- Generate the data using the CreateDataset python notebook
- Change in the config and dataset name in Baseline notebook to run and save the baseline model.
- Run the commonsense infused model in the GCNModel notebook
@inproceedings{basu-roy-chowdhury-chaturvedi-2021-commonsense,
title = "Does Commonsense help in detecting Sarcasm?",
author = "Basu Roy Chowdhury, Somnath and
Chaturvedi, Snigdha",
booktitle = "Proceedings of the Second Workshop on Insights from Negative Results in NLP",
month = nov,
year = "2021",
address = "Online and Punta Cana, Dominican Republic",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2021.insights-1.2",
doi = "10.18653/v1/2021.insights-1.2",
}