Named entity recognition (NER) is the widely studied task consisting in identifying text spans that denote named entities such as person, location and organization names, to name the most important types. Such text spans are called named entity mentions. In NER, mentions are generally not only identified, but also classified according to a more or less fine-grained ontology, thereby allowing for instance to distinguish between the telecommunication company Orange and the town Orange in southern France (amongst others). (Ortiz Suárez et al. 2020, p. 4631)
FTB French Tree Bank, Universal Dependencies UD
FI-score 90,25.
Précision/Rappel
Le F1-score est une métrique de classification qui mesure la capacité d’un modèle à bien prédire les individus positifs, tant en termes de precision (taux de prédictions positives correctes) qu’en termes de recall (taux de positifs correctement prédits). Il correspond en effet à la moyenne harmonique de ces indicateurs, qui doivent tous deux être élevés pour que le F1-score le soit aussi.
https://universaldependencies.org/treebanks/fr_ftb/
- Ortiz Suárez, Pedro Javier, Yoann Dupont, Benjamin Muller, Laurent Romary, et Benoît Sagot. 2020. « Establishing a New State-of-the-Art for French Named Entity Recognition ». In Proceedings of the Twelfth Language Resources and Evaluation Conference, 4631‑38. Marseille, France: European Language Resources Association. https://aclanthology.org/2020.lrec-1.569.
- Liu, Weijie, Peng Zhou, Zhe Zhao, Zhiruo Wang, Qi Ju, Haotang Deng, et Ping Wang. 2019. « K-BERT: Enabling Language Representation with Knowledge Graph ». arXiv. https://doi.org/10.48550/arXiv.1909.07606.