DaDoEval / README.md
MattiaSangermano's picture
Added references
7a78eb7 verified
metadata
dataset_info:
  features:
    - name: id
      dtype: string
    - name: year
      dtype: int64
    - name: class
      dtype:
        class_label:
          names:
            '0': Beginning of political activity
            '1': Building the Italian Republic
            '2': From fascism to the Italian Republic
            '3': Habsburg years
            '4': Internal exile
    - name: year_range
      dtype:
        class_label:
          names:
            '0': 1901-1905
            '1': 1906-1910
            '2': 1911-1915
            '3': 1916-1920
            '4': 1921-1925
            '5': 1926-1930
            '6': 1931-1935
            '7': 1936-1940
            '8': 1941-1945
            '9': 1946-1950
            '10': 1951-1955
    - name: text
      dtype: string
  splits:
    - name: train
      num_bytes: 16007781
      num_examples: 2210
    - name: test
      num_bytes: 4513601
      num_examples: 549
    - name: test_ood
      num_bytes: 186150
      num_examples: 100
  download_size: 12416424
  dataset_size: 20707532
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: test
        path: data/test-*
      - split: test_ood
        path: data/test_ood-*
license: cc-by-nc-sa-4.0
task_categories:
  - text-classification
  - text-retrieval
language:
  - it
size_categories:
  - 1K<n<10K

DaDoEval

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

Overview

The DaDoEval dataset is a curated collection of 2,759 documents authored by Alcide De Gasperi, spanning the period from 1901 to 1954. Each document in the dataset is manually tagged with its date of issue. This dataset is designed for the temporal classification task, specifically focused on assigning a temporal span to documents. The corpus is divided into a training set and an in-domain test set to facilitate supervised learning approaches. In addition to the primary dataset, a cross-genre out-of-domain test set consisting of approximately 100 letters from the Epistolario project is included.

Sub-tasks:

  1. Coarse-grained Classification: Assign documents to one of five historical periods in De Gasperi's life.
  2. Fine-grained Classification: Assign documents to a 5-year temporal slice.
  3. Year-based Classification: Assign documents to their exact year of publication.

Historical Periods:

The following table outlines the periods defined by historians for the coarse-grained classification task:

Period Years
Habsburg years 1901-1918
Beginning of political activity 1919-1926
Internal exile 1927-1942
From fascism to the Italian Republic 1943-1947
Building the Italian Republic 1948-1954

Dataset Structure:

  • id: A unique identifier for each document.
  • text: The full text content of the document.
  • year: The exact year the document was issued.
  • class: The historical period the document belongs to.
  • year_range: A 5-year temporal slice indicating the period the document falls into.

Citation

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

@inproceedings{menini2020dadoeval,
  title={DaDoEval@ EVALITA 2020: Same-genre and cross-genre dating of historical documents},
  author={Menini, Stefano and Moretti, Giovanni and Sprugnoli, Rachele and Tonelli, Sara and others},
  booktitle={Proceedings of the Seventh Evaluation Campaign of Natural Language Processing and Speech Tools for Italian. Final Workshop (EVALITA 2020)},
  pages={391--397},
  year={2020},
  organization={Accademia University Press}
}