Papers by Santiago Góngora
Jojajovai: A Parallel Guarani-Spanish Corpus for MT Benchmarking (2022.lrec-1)
Copied to clipboard
Luis Chiruzzo, Santiago Góngora, Aldo Alvarez, Gustavo Giménez-Lugo, Marvin Agüero-Torales, Yliana Rodríguez
| Challenge: | a corpus of Guarani-Spanish text is presented that is aligned at sentence level . the long history of language contact between Guaran and Spanish in South America has resulted in many interesting language varieties . |
| Approach: | They propose to align Guarani-Spanish text at sentence level with 30,000 sentence pairs and a test set. |
| Outcome: | The proposed corpus contains about 30,000 sentence pairs and is structured as a collection of subsets from different sources, further split into training, development and test sets. |
Grammar-based Data Augmentation for Low-Resource Languages: The Case of Guarani-Spanish Neural Machine Translation (2024.naacl-long)
Copied to clipboard
Agustín Lucas, Alexis Baladón, Victoria Pardiñas, Marvin Agüero-Torales, Santiago Góngora, Luis Chiruzzo
| Challenge: | Low-resource languages suffer from a vicious circle: data is needed to build tools, but available text is scarce. |
| Approach: | They propose to use a grammar-based system to generate Spanish text and syntactically transfer it to Guarani to boost its performance. |
| Outcome: | The proposed system outperforms existing models by pretraining models with synthetic text. |