Skip to content

Latest commit

 

History

History
61 lines (43 loc) · 3.47 KB

README.md

File metadata and controls

61 lines (43 loc) · 3.47 KB
Raw

ReactionGIF

ReactionGIF is a unique, first-of-its-kind dataset of 30K tweets and their GIF reactions.

To find out more about ReactionGIF, check out our ACL 2021 paper:

Use this repository to download ReactionGIF. The repository includes the following data file with the tweet information:

  • ReactionGIF.ids.json with original tweet IDs in jsonlines format.

Each record in the file includes the following fields:

  • idx record number (note: record numbers are not sequential)
  • original_id the tweet ID of the original tweet which contains the eliciting text
  • reply_id the tweet ID of the reply tweet which contains the reaction GIF
  • label the reaction category

To comply with Twitter's ToS and Developer Agreement and Policy, the dataset includes only the tweet IDs. To fetch the original tweets' texts, you can write your own script, or you can use our own, easy-to-use script. To use our own script, follow these steps:

  1. Clone the repository or download the files ReactionGIF.ids.json, credentials-example.py, and fetch-tweets.py
  2. Install the latest version of Tweepy:
pip3 install tweepy
  1. Rename our credentials-example.py to credentials.py
mv credentials-example.py credentials.py
  1. Add your Twitter API credentials by editing credentials.py:
vim credentials.py
  1. Run the script:
python3 fetch-tweets.py [--gifs]

The script will fetch the tweet texts and add a new text field. If you turn on the gifs flag, the script will also add a reply field which will include the link to the GIF. The new dataset will be saved to:

  • ReactionGIF.json

Citation

If you use our dataset, kindly cite the paper using the following BibTex entry:

@inproceedings{shmueli-etal-2021-happy,
    title = "Happy Dance, Slow Clap: {Using} Reaction {GIFs} to Predict Induced Affect on {Twitter}",
    author = "Shmueli, Boaz  and
      Ray, Soumya  and
      Ku, Lun-Wei",
    booktitle = "Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 2: Short Papers)",
    month = aug,
    year = "2021",
    address = "Online",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2021.acl-short.50",
    doi = "10.18653/v1/2021.acl-short.50",
    pages = "395--401",
    abstract = "Datasets with induced emotion labels are scarce but of utmost importance for many NLP tasks. We present a new, automated method for collecting texts along with their induced reaction labels. The method exploits the online use of reaction GIFs, which capture complex affective states. We show how to augment the data with induced emotion and induced sentiment labels. We use our method to create and publish ReactionGIF, a first-of-its-kind affective dataset of 30K tweets. We provide baselines for three new tasks, including induced sentiment prediction and multilabel classification of induced emotions. Our method and dataset open new research opportunities in emotion detection and affective computing.",
}