Datasets:
Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,141 @@
|
|
| 1 |
---
|
| 2 |
-
|
| 3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
+
YAML tags: null
|
| 3 |
+
language:
|
| 4 |
+
- es
|
| 5 |
+
- ast
|
| 6 |
+
pretty_name: ES-AST Parallel Corpus
|
| 7 |
+
task_categories:
|
| 8 |
+
- translation
|
| 9 |
+
size_categories:
|
| 10 |
+
- size category
|
| 11 |
+
---
|
| 12 |
+
|
| 13 |
+
# Dataset Card for ES-AST Parallel Corpus
|
| 14 |
+
|
| 15 |
+
## Dataset Description
|
| 16 |
+
|
| 17 |
+
- **Point of Contact:** langtech@bsc.es
|
| 18 |
+
|
| 19 |
+
|
| 20 |
+
### Dataset Summary
|
| 21 |
+
|
| 22 |
+
The ES-AN Parallel Corpus is a Spanish-Asturian dataset created to support the use of under-resourced languages from Spain,
|
| 23 |
+
such as Asturian, in NLP tasks, specifically Machine Translation.
|
| 24 |
+
|
| 25 |
+
|
| 26 |
+
### Supported Tasks and Leaderboards
|
| 27 |
+
|
| 28 |
+
The dataset can be used to train Bilingual Machine Translation models between Asturian and Spanish in any direction, as well as Multilingual Machine Translation models.
|
| 29 |
+
|
| 30 |
+
### Languages
|
| 31 |
+
|
| 32 |
+
The sentences included in the dataset are in Spanish (ES) and Asturian (AST).
|
| 33 |
+
|
| 34 |
+
## Dataset Structure
|
| 35 |
+
|
| 36 |
+
### Data Instances
|
| 37 |
+
|
| 38 |
+
Two separate txt files are provided:
|
| 39 |
+
|
| 40 |
+
- es-ast_corpus.es
|
| 41 |
+
- es-ast_corpus.ast
|
| 42 |
+
|
| 43 |
+
The dataset is additionally provided in parquet format: es-ast_corpus.parquet.
|
| 44 |
+
|
| 45 |
+
The parquet file contains two columns of parallel text obtained from the two original text files.
|
| 46 |
+
Each row in the file represents a pair of parallel sentences in the two languages of the dataset.
|
| 47 |
+
|
| 48 |
+
|
| 49 |
+
### Data Fields
|
| 50 |
+
|
| 51 |
+
[N/A]
|
| 52 |
+
|
| 53 |
+
### Data Splits
|
| 54 |
+
|
| 55 |
+
The dataset contains a single split: `train`.
|
| 56 |
+
|
| 57 |
+
## Dataset Creation
|
| 58 |
+
|
| 59 |
+
### Curation Rationale
|
| 60 |
+
|
| 61 |
+
This dataset is aimed at promoting the development of Machine Translation between Spanish and under-resourced languages from Spain, specifically Asturian.
|
| 62 |
+
|
| 63 |
+
### Source Data
|
| 64 |
+
|
| 65 |
+
#### Initial Data Collection and Normalization
|
| 66 |
+
|
| 67 |
+
This dataset was created as part of the participation of Language Technologies Unit at BSC in the WMT24 Shared Task:
|
| 68 |
+
[Translation into Low-Resource Languages of Spain](https://www2.statmt.org/wmt24/romance-task.html).
|
| 69 |
+
The corpus is the result of a thorough cleaning and preprocessing, as described in detail in the paper "Training and Fine-Tuning
|
| 70 |
+
NMT Models for Low-Resource Languages using Apertium-Based Synthetic Corpora" (link to be added as soon as published).
|
| 71 |
+
As no filtering based on alignment score was applied, the dataset may contain poorly aligned sentences.
|
| 72 |
+
|
| 73 |
+
This dataset aggregates both synthetic and authentic data.
|
| 74 |
+
The authentic parallel data come from [OPUS](https://opus.nlpl.eu/) (Spanish-Asturian), while the synthetic data consist of Spanish translations generated from the Asturian monolingual corpus of the [PILAR](https://github.com/transducens/PILAR) dataset.
|
| 75 |
+
To create the synthetic Spanish we used the rule-based [Apertium](https://www.apertium.org/) translator.
|
| 76 |
+
|
| 77 |
+
#### Who are the source language producers?
|
| 78 |
+
|
| 79 |
+
[Opus](https://opus.nlpl.eu/)
|
| 80 |
+
|
| 81 |
+
[PILAR](https://github.com/transducens/PILAR)
|
| 82 |
+
|
| 83 |
+
[WMT24](https://www2.statmt.org/wmt24/romance-task.html)
|
| 84 |
+
|
| 85 |
+
|
| 86 |
+
### Annotations
|
| 87 |
+
|
| 88 |
+
#### Annotation process
|
| 89 |
+
|
| 90 |
+
The dataset does not contain any annotations.
|
| 91 |
+
|
| 92 |
+
#### Who are the annotators?
|
| 93 |
+
|
| 94 |
+
[N/A]
|
| 95 |
+
|
| 96 |
+
### Personal and Sensitive Information
|
| 97 |
+
|
| 98 |
+
Given that this dataset is partly derived from pre-existing datasets that may contain crawled data, and that no specific anonymisation process has been applied,
|
| 99 |
+
personal and sensitive information may be present in the data. This needs to be considered when using the data for training models.
|
| 100 |
+
|
| 101 |
+
## Considerations for Using the Data
|
| 102 |
+
|
| 103 |
+
### Social Impact of Dataset
|
| 104 |
+
|
| 105 |
+
By providing this resource, we intend to promote the use of Asturian across NLP tasks,
|
| 106 |
+
thereby improving the accessibility and visibility of the Asturian language.
|
| 107 |
+
|
| 108 |
+
### Discussion of Biases
|
| 109 |
+
|
| 110 |
+
No specific bias mitigation strategies were applied to this dataset.
|
| 111 |
+
Inherent biases may exist within the data.
|
| 112 |
+
|
| 113 |
+
### Other Known Limitations
|
| 114 |
+
|
| 115 |
+
The dataset contains data of a general domain. Applications of this dataset in more specific domains such as biomedical, legal etc. would be of limited use.
|
| 116 |
+
|
| 117 |
+
## Additional Information
|
| 118 |
+
|
| 119 |
+
### Dataset Curators
|
| 120 |
+
|
| 121 |
+
Language Technologies Unit at the Barcelona Supercomputing Center (langtech@bsc.es).
|
| 122 |
+
|
| 123 |
+
This work is funded by the Ministerio para la Transformación Digital y de la Función Pública and Plan de Recuperación,
|
| 124 |
+
Transformación y Resiliencia - Funded by EU – NextGenerationEU within the framework of the project ILENIA with reference
|
| 125 |
+
2022/TL22/00215337, 2022/TL22/00215336, 2022/TL22/00215335, 2022/TL22/00215334.
|
| 126 |
+
|
| 127 |
+
The publication is part of the project PID2021-123988OB-C33, funded by MCIN/AEI/10.13039/501100011033/FEDER, EU.
|
| 128 |
+
|
| 129 |
+
|
| 130 |
+
### Licensing Information
|
| 131 |
+
|
| 132 |
+
This work is licensed under a [Attribution-NonCommercial-ShareAlike 4.0 International](https://creativecommons.org/licenses/by-nc-sa/4.0/)
|
| 133 |
+
due to licence restrictions on part of the original data.
|
| 134 |
+
|
| 135 |
+
### Citation Information
|
| 136 |
+
|
| 137 |
+
[N/A]
|
| 138 |
+
|
| 139 |
+
### Contributions
|
| 140 |
+
|
| 141 |
+
[N/A]
|