|
|
--- |
|
|
configs: |
|
|
- config_name: english |
|
|
data_files: NunavutHansard.en.jsonl |
|
|
- config_name: inuktitut |
|
|
data_files: NunavutHansard.iu.rom.jsonl |
|
|
- config_name: inuktitut_syllabics |
|
|
data_files: NunavutHansard.iu.syl.jsonl |
|
|
license: cc-by-4.0 |
|
|
task_categories: |
|
|
- translation |
|
|
language: |
|
|
- en |
|
|
- iu |
|
|
tags: |
|
|
- indigenous |
|
|
- legal |
|
|
pretty_name: Nunavut Hansard Inuktitut–English Parallel Corpus 3.0++ |
|
|
size_categories: |
|
|
- 1M<n<10M |
|
|
--- |
|
|
|
|
|
This repo contains an processed version of the [Nunavut Hansard Inuktitut–English Parallel Corpus 3.0](https://doi.org/10.4224/40001819) dataset. |
|
|
The dataset has been deduped and blank lines removed. Further, both corpora have lemma and part of speech information. |
|
|
For English, this information has been added via [SpaCy](https://spacy.io)'s `en_core_web_trf` model. |
|
|
For Inuktitut, this information comes from automatic neural and rule-based morphological analyses of words courtesy of [Jeffrey Micher](https://www.cs.cmu.edu/afs/cs/project/llab/www/inuk.html). |
|
|
Inuktitut additionally contains a translation of the lemma, where possible, which comes from the Uqailaut analyzer source. |
