bourdoiscatie's picture
Create README.md
1536dfe
metadata
language:
  - fr
license:
  - unknown
size_categories:
  - 10K<n<100K
task_categories:
  - token-classification
tags:
  - ner

wikiner_fr_prompt_ner

Summary

wikiner_fr_prompt_ner is a subset of the Dataset of French Prompts (DFP). It contains X rows that can be used for a name entity recognition task. The original data (without prompts) comes from the dataset wikiner by Nothman et al. where only the French part has been kept. A list of prompts (see below) was then applied in order to build the input and target columns and thus obtain the same format as the xP3 dataset by Muennighoff et al.

Prompts used

List

21 prompts were created for this dataset. The logic applied consists in proposing prompts in the indicative tense, in the form of tutoiement and in the form of vouvoiement.

'Extraire les classes des mots du texte suivant : '+text,
'Extrais les entitées nommées du texte suivant : '+text,
'Extrayez les entitées nommées du texte suivant : '+text,
'Isoler les entitées nommées du texte suivant : '+text,
'Isole les entitées nommées du texte suivant : '+text,
'Isolez les entitées nommées du texte suivant : '+text,
'Dégager des entitées nommées dans le texte : '+text,
'Dégage des entitées nommées dans le texte : '+text,
'Dégagez des entitées nommées dans le texte : '+text,
'Générer des entitées nommées issues du texte suivant : '+text,
'Génère des entitées nommées issues du texte suivant : '+text,
'Générez des entitées nommées issues du texte suivant : '+text,
'Trouver les entitées nommées du texte : '+text,
'Trouve les entitées nommées du texte : '+text,
'Trouvez les entitées nommées du texte : '+text,
'Repérer les entitées nommées présentes dans le texte suivant : '+text,
'Repère les entitées nommées présentes dans le texte suivant : '+text,
'Repérez les entitées nommées présentes dans le texte suivant : '+text,
'Indiquer les entitées nommées du texte :'+text,
'Indique les entitées nommées du texte : '+text,
'Indiquez les entitées nommées du texte : '+text

Features used in the prompts

In the prompt list above, text and targets have been constructed from:

wikiner_fr = load_dataset('Jean-Baptiste/wikiner_fr')  
wikiner_fr['train']['tokens'] = list(map(lambda i: ' '.join(wikiner_fr['train']['tokens'][i]), range(len(wikiner_fr['train']['tokens']))))
wikiner_fr['train']['ner_tags'] = list(map(lambda x: x.replace("[","").replace("]","").replace('0','O').replace('1','LOC').replace('2','PER').replace('3','O').replace('4','ORG'), map(str, wikiner_fr['train']['ner_tags'])))

Splits

  • train with X samples
  • dev with Y samples
  • test with Z samples

How to use?

from datasets import load_dataset
dataset = load_dataset("CATIE-AQ/wikiner_fr_prompt_ner")

Citation

Original data

@article{NOTHMAN2013151, title = {Learning multilingual named entity recognition from Wikipedia}, journal = {Artificial Intelligence}, volume = {194}, pages = {151-175}, year = {2013}, note = {Artificial Intelligence, Wikipedia and Semi-Structured Resources}, issn = {0004-3702}, doi = {https://doi.org/10.1016/j.artint.2012.03.006}, url = {https://www.sciencedirect.com/science/article/pii/S0004370212000276}, author = {Joel Nothman and Nicky Ringland and Will Radford and Tara Murphy and James R. Curran}, keywords = {Named entity recognition, Information extraction, Wikipedia, Semi-structured resources, Annotated corpora, Semi-supervised learning}, abstract = {We automatically create enormous, free and multilingual silver-standard training annotations for named entity recognition (ner) by exploiting the text and structure of Wikipedia. Most ner systems rely on statistical models of annotated data to identify and classify names of people, locations and organisations in text. This dependence on expensive annotation is the knowledge bottleneck our work overcomes. We first classify each Wikipedia article into named entity (ne) types, training and evaluating on 7200 manually-labelled Wikipedia articles across nine languages. Our cross-lingual approach achieves up to 95% accuracy. We transform the links between articles into ne annotations by projecting the target articleʼs classifications onto the anchor text. This approach yields reasonable annotations, but does not immediately compete with existing gold-standard data. By inferring additional links and heuristically tweaking the Wikipedia corpora, we better align our automatic annotations to gold standards. We annotate millions of words in nine languages, evaluating English, German, Spanish, Dutch and Russian Wikipedia-trained models against conll shared task data and other gold-standard corpora. Our approach outperforms other approaches to automatic ne annotation (Richman and Schone, 2008 [61], Mika et al., 2008 [46]) competes with gold-standard training when tested on an evaluation corpus from a different source; and performs 10% better than newswire-trained models on manually-annotated Wikipedia text.} }

This Dataset

License

Unknow