README.md

metadata

dataset_info:
  features:
    - name: unique_ID
      dtype: large_string
    - name: ID
      dtype: large_string
    - name: text_ID
      dtype: int64
    - name: subcorpus
      dtype: large_string
    - name: author
      dtype: large_string
    - name: date
      dtype: timestamp[ms]
    - name: Text
      dtype: large_string
    - name: text_tokenised
      dtype: large_string
    - name: token_count
      dtype: int64
    - name: similar_IDs
      list: string
    - name: annotation_prompt
      dtype: large_string
    - name: annotation_model
      dtype: large_string
    - name: annotation_temperature
      dtype: float64
    - name: annotation_top_p
      dtype: float64
    - name: annotation_max_model_len
      dtype: int64
    - name: synthetic_labels
      list:
        - name: Aspekt
          dtype: string
        - name: Sentimentbegriff
          dtype: string
        - name: Valenz
          dtype: string
        - name: Erregung
          dtype: string
    - name: messages
      list:
        - name: role
          dtype: string
        - name: content
          dtype: string
  splits:
    - name: train
      num_bytes: 28892954
      num_examples: 3934
    - name: test
      num_bytes: 3434382
      num_examples: 463
    - name: dev
      num_bytes: 1751377
      num_examples: 232
  download_size: 12609414
  dataset_size: 35883436
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: test
        path: data/test-*
      - split: dev
        path: data/dev-*
language:
  - de
size_categories:
  - 1K<n<10K

AfD-CCC

A corpus of short texts and parliamentary speech transcripts by the German political party "Alternative für Deutschland" (AfD) addressing the topic of climate change (CC). The collection process and some exemplary usecases of the corpus are outlined in the associated paper and master's thesis. All data sources are clearly demarked in the following section.

Data Sources

Documents are taken from the original repository at github.com/discourse-lab/afd-ccc.

Manfred Stede and Ronja Memminger. 2025. AfD-CCC: Analyzing the Climate Change Discourse of a German Right-wing Political Party. In Proceedings of the Fourth Workshop on NLP for Positive Impact (NLP4PI), pages 163–174, Vienna, Austria. Association for Computational Linguistics.

• German Parliament speeches (2017-2022): Extracted from OpenDiscourse (Richter et al., 2020)
• German Parliament speeches (2022-2025): Retrieved from Government records via the DIP API.
• European Parliament speeches (2014-2024): Extracted from the ParlLawSpeech dataset, Version 1.0 (Schwabach et al., 2025)
• Telegram messages (2019-2025): Scraped from public channels using FROG (Priming & Fröschl, 2024)
• Press releases (2017-2021): Extracted from press releases made available by Schaefer et al. (2023)
• Tweets (2017-2022): From Lasser et al. (2022). Due to Twitter privacy regulations, only tweet IDs can be given. (Excluded due to missing documents.)

Corpus Statistics

Text Domain	Texts	Tokens	Annotated texts	Total # synthetic labels
German Parl.	2,284	220,894	2,197	12,332
European Parl.	668	48,025	659	3,017
Press releases	911	85,356	896	5,012
Twitter	1,880	54,592	-	-
Telegram	766	41,487	748	3,297
Total	4,629	395,762	4,500	23,658