|
--- |
|
license: mit |
|
--- |
|
|
|
|
|
. |
|
# Spanish truecasing model |
|
|
|
This is a Spanish truecasing-model that works with the <b>Dalton Fury</b> Python project: |
|
|
|
https://github.com/daltonfury42/truecase |
|
|
|
You can install it here: |
|
|
|
https://pypi.org/project/truecase/ |
|
|
|
## Quick start |
|
|
|
To use the Spanish model use the TrueCase.py file uploaded to this repository |
|
|
|
https://huggingface.co/HURIDOCS/spanish_truecasing/blob/main/TrueCaser.py |
|
|
|
Install the requirements: |
|
|
|
pip install nltk |
|
|
|
And ready to work: |
|
|
|
from TrueCaser import TrueCaser |
|
|
|
model_path = "spanish.dist" |
|
spanish_truecasing = TrueCaser(model_path) |
|
|
|
text = 'informe no.78/08. petición 785-05 admisibilidad. vicente arturo villanueva ortega y otros.' |
|
print(spanish_truecasing.get_true_case(text)) |
|
|
|
|
|
## Notes |
|
|
|
The model was trained with the Europarl dataset that contains transcriptions of the European Parliament discusions: |
|
|
|
https://www.statmt.org/europarl/ |
|
Europarl: A Parallel Corpus for Statistical Machine Translation, Philipp Koehn, MT Summit 2005 |
|
|
|
Using huggingface load_dataset: |
|
|
|
europarl = load_dataset('large_spanish_corpus', name='Europarl') |
