README.md

metadata

license: cc0-1.0
task_categories:
  - text-classification
  - token-classification
language:
  - en
multilinguality:
  - monolingual
size_categories:
  - 1K<n<10K
tags:
  - causality
pretty_name: Causal News Corpus (CNC)
configs:
  - config_name: causality detection
    data_files:
      - split: train
        path: causality-detection/train.parquet
      - split: test
        path: causality-detection/test.parquet
    features:
      - name: index
        dtype: string
      - name: text
        dtype: string
      - name: label
        dtype:
          class_label:
            names:
              '0': uncausal
              '1': causal
  - config_name: causal candidate extraction
    data_files:
      - split: train
        path: causal-candidate-extraction/train.parquet
      - split: test
        path: causal-candidate-extraction/test.parquet
    features:
      - name: index
        dtype: string
      - name: text
        dtype: string
      - name: entity
        sequence:
          sequence: int32
  - config_name: causality identification
    data_files:
      - split: train
        path: causality-identification/train.parquet
      - split: test
        path: causality-identification/test.parquet
    features:
      - name: index
        dtype: string
      - name: text
        dtype: string
      - name: relations
        list:
          - name: relationship
            dtype:
              class_label:
                names:
                  '0': no-rel
                  '1': causal
          - name: first
            dtype: string
          - name: second
            dtype: string
train-eval-index:
  - config: causality detection
    task: text-classification
    task_id: text_classification
    splits:
      train_split: train
      eval_split: test
    col_mapping:
      text: text
      label: label
    metrics:
      - type: accuracy
      - type: precision
      - type: recall
      - type: f1
  - config: causal candidate extraction
    task: token-classification
    task_id: token_classification
    splits:
      train_split: train
      eval_split: test
    metrics:
      - type: accuracy
      - type: precision
      - type: recall
      - type: f1
  - config: causality identification
    task: text-classification
    task_id: text_classification
    splits:
      train_split: train
      eval_split: test
    metrics:
      - type: accuracy
      - type: precision
      - type: recall
      - type: f1

This repository integrates the original 2022 "V1" release of the Causal News Corpus (CNC) into hf datasets. Please find the original dataset here. This is the OLDER, much more sparsely span-annotated release (183 causal relations, vs. 2257 in CNCv2, the "V2"/RECESS release the maintainers now recommend using) — kept as its own separate dataset rather than silently overwritten, so both remain available for comparison. Please see the citations at the end of this README.

Dataset Description

• Repository: https://github.com/tanfiona/CausalNewsCorpus
• Paper: The Causal News Corpus: Annotating Causal Relations in Event Sentences

Usage

Causality Detection

from datasets import load_dataset
dataset = load_dataset("thagen/CausalNewsCorpus", "causality detection")

Causal Candidate Extraction

from datasets import load_dataset
dataset = load_dataset("thagen/CausalNewsCorpus", "causal candidate extraction")

Causality Identification

from datasets import load_dataset
dataset = load_dataset("thagen/CausalNewsCorpus", "causality identification")

Citations

The Causal News Corpus paper by Tan et al., 2022:

@inproceedings{tan:2022,
  title = {The Causal News Corpus: Annotating Causal Relations in Event Sentences},
  booktitle = {Proceedings of the 13th Language Resources and Evaluation Conference},
  author = {Tan, Fiona Anting and Ng, See-Kiong and Ong, Alifia Reina},
  year = {2022},
  pages = {2298--2310},
  publisher = {European Language Resources Association}
}