metadata

language:
  - de
multilinguality:
  - monolingual
license: cc-by-4.0
pretty_name: SB10k
task_categories:
  - text-classification
tags:
  - Sentiment Analysis
size_categories:
  - 1K<n<10K
configs:
  - config_name: default
    sep: "\t"
    column_names:
      - ID
      - Sentiment
      - Text
      - Normalized
      - POS-Tags
      - Dependency Labels
      - additional Annotations
    data_files:
      - split: train
        path: train.tsv
      - split: test
        path: test.tsv
      - split: dev
        path: dev.tsv

A Twitter corpus and benchmark resources for german sentiment analysis

Source

The data is a snapshot from the SB10k Dataset. The snapshot was made by Oliver Guhr.

Citation Information

Paper

@inproceedings{cieliebak2017twitter,
  title={A twitter corpus and benchmark resources for german sentiment analysis},
  author={Cieliebak, Mark and Deriu, Jan Milan and Egger, Dominic and Uzdilli, Fatih},
  booktitle={5th International Workshop on Natural Language Processing for Social Media, Boston MA, USA, 11 December 2017},
  pages={45--51},
  year={2017},
  organization={Association for Computational Linguistics}
}