proxectonos/UD_Galician-PUD

Galician treebank

Dataset containing 1000 manually reviewed sentences aligned with the other available PUD treebanks. The sentences are taken from news journals (sentence id starts with 'n') and from Wikipedia (sentence id starts with 'w'), with no more than a few sentences per document, which are randomly selected. The next two digits in the sentence id encode its original language: the first 750 sentences are originally English (01) and the remaining 250 sentences are originally German (02), French (03), Italian (04) or Spanish (05) and they were translated to other languages via English.

Its labeled as a test set in its entirety and was annotated with a two step approach: it was first annotated using state-of-the-art Galician NLP and then reviewed by two experts, with a high inter-annotator agreement.

Acknowledgments

The first version of this dataset was compiled within the Nós Project, funded by the Ministerio para la Transformación Digital y de la Función Pública - Funded by EU – NextGenerationEU within the framework of the project ILENIA with reference 2022/TL22/00215336. This work was also funded by the Ministerio para la Transformación Digital y de la Función Pública - Funded by EU – NextGenerationEU within the framework of the project Desarrollo de Modelos ALIA. Esta publicación del proyecto Desarrollo de Modelos ALIA está financiada por el Ministerio para la Transformación Digital y de la Función Pública y por el Plan de Recuperación, Transformación y Resiliencia – Financiado por la Unión Europea – NextGenerationEU.

Citation:

Xulia Sánchez, Albina Sarymsakova, Laura Castro, Marcos García. Increasing manually annotated resources for Galician: the Parallel Universal Dependencies Treebank.