Source code for the paper "Hashtag-Guided Low-Resource Tweet Classification"
You can install all the dependencies by source install.sh.
All the data we used in our paper is in data folder, which is based on the data from TweetEval.
The file hashtag_generation.py trains (and saves) the hashtag generation models, while the file hashtag_generation_inference.py uses the saved models to generate the hashtags. Lastly, the file classification.py does the classification and evaluation for the TweetEval tasks.
If you find this repository useful, you may cite our paper as:
@inproceedings{diao-etal-2023-hashtation,
title={Hashtag-Guided Low-Resource Tweet Classification},
author={Shizhe Diao and Sedrick Scott Keh and Liangming Pan and Zhiliang Tian and Yan Song and Tong Zhang},
year={2023},
booktitle={The Web Conference 2023}
}
For help or issues using this package, please submit a GitHub issue.
For personal communication related to this package, please contact Shizhe Diao (sdiaoaa@connect.ust.hk) and Sedrick Scott Keh (skeh@cs.cmu.edu).