: Vocabulaire du TAL: training data

NLP resources and evaluation > data set > training data

training data

It seems important to point out the fact that the average sentence length in the training sets is much shorter than in the other sets is because in the cited workshop the training sets were restricted to sentences with a maximum length of 40 words whereas the rest of sets did not have this restriction. (González-Rubio, Sanchis-Trilles, Juan & Casacuberta, 2008)
The corresponding training data includes a high number of examples which vary in terms of their validity label. (Heinisch, Plenz, Opitz, Frank & Cimiano, 2022)
The training split includes 10009 videos with 4917 videos allocated for testing. (Jian & Wang, 2023)
To prepare training data for such a system we begin with a bilingual text that has been automatically processed into segment pairs. (Wang, May, Knight & Marcu, 2010)
We use 20% of the training dataset as validation set. (Kumar, Sethi, Akhtar, Ekbal, Biemann & Bhattacharyya, 2017)

http://data.loterre.fr/ark:/67375/8LP-QL538VQK-1

RDF/XML TURTLE JSON-LD Dernière modification le 27/05/2024

Loterre