metadata

license: cc-by-nc-sa-4.0
dataset_info:
  features:
    - name: tweet_id
      dtype: int64
    - name: text
      dtype: string
    - name: label
      dtype:
        class_label:
          names:
            '0': AGAINST
            '1': FAVOR
            '2': NONE
  splits:
    - name: train
      num_bytes: 417810
      num_examples: 2132
    - name: test
      num_bytes: 215749
      num_examples: 1110
  download_size: 444162
  dataset_size: 633559
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: test
        path: data/test-*
task_categories:
  - text-classification
language:
  - it
pretty_name: ff
size_categories:
  - 1K<n<10K

SardiStance

Disclaimer: This dataset is not the official SardiStance repository from EVALITA. For the official dataset and more information, please visit the EVALITA SardiStance page and the SardiStance repository

SardiStance is a unique dataset designed for the task of stance detection in Italian tweets. It consists of tweets related to the Sardines movement, providing a valuable resource for researchers and practitioners in the field of NLP. This dataset was curated for the EVALITA 2020 competition, aiming to advance the automatic detection of stance in social media text.

Dataset Description

The SardiStance dataset contains tweets written in Italian, specifically focusing on the Sardines movement. The primary goal is to classify the stance of each tweet into one of three categories: AGAINST, FAVOR or NONE towards the Sardines movement. The dataset includes a variety of tweets, capturing different opinions and sentiments expressed by users on this political movement.

Data Format and Annotations

The entire SardiStance dataset consists of 3,242 instances, divided into training and test sets. The training set contains 2,132 tweets, while the test set includes 1,110 tweets. The dataset was collected by retrieving tweets containing keywords related to the Sardines movement and their associated hashtags. Retweets, quotes, replies, and tweets containing URLs or extended entities (videos, photos, GIFs) were excluded to ensure data quality.

The dataset is composed of the following columns:

tweet_id: the id of the sample
text: The dataset includes the full text of each tweet.
label: Each tweet is annotated with one of three stance labels: AGAINST, FAVOR or NONE.

Citations:

If you use this dataset, please cite the original authors:

@inproceedings{cignarella2020sardistance,
  title={Sardistance@ evalita2020: Overview of the task on stance detection in italian tweets},
  author={Cignarella, Alessandra Teresa and Lai, Mirko and Bosco, Cristina and Patti, Viviana and Rosso, Paolo and others},
  booktitle={CEUR WORKSHOP PROCEEDINGS},
  pages={1--10},
  year={2020},
  organization={Ceur}
}