Hugging Face's logo

Datasets:
undertheseanlp
/
UDD-1

Tasks:
Token Classification
Modalities:
Document
Text
Formats:
parquet
Sub-tasks:
parsing
part-of-speech
lemmatization
Languages:
Vietnamese
Size:
10K - 100K
Tags:
universal-dependencies
vietnamese
nlp
dependency-parsing
pos-tagging
Libraries:
Datasets
pandas
Polars
License:
Dataset card Data Studio Files
xet
Community
1
UDD-1 / README.md
rain1024's picture
rain1024
Update dataset card: add UVN-1 news domain (20k sentences total)
21c6a4e verified
preview code
|
raw
history blame contribute delete
2.99 kB
metadata 
license: cc-by-sa-4.0
task_categories:
  - token-classification
task_ids:
  - parsing
  - part-of-speech
  - lemmatization
language:
  - vi
annotations_creators:
  - machine-generated
language_creators:
  - found
multilinguality:
  - monolingual
source_datasets:
  - undertheseanlp/UTS_VLC
  - undertheseanlp/UVN-1
size_categories:
  - 10K<n<100K
tags:
  - universal-dependencies
  - vietnamese
  - nlp
  - dependency-parsing
  - pos-tagging

UDD-1: Universal Dependency Dataset for Vietnamese

Dataset Description

Vietnamese Universal Dependency dataset created by Underthesea NLP. This dataset follows the Universal Dependencies annotation guidelines.

Dataset Summary

  • Language: Vietnamese (vi)
  • Version: 1.1
  • Domains: ⚖️ Legal (Laws) + 📰 News
  • Total Sentences: 20,000
  • Total Tokens: ~453,551
  • Annotation: Machine-generated using Underthesea NLP toolkit

Data Sources

Source Domain Sentences Tokens
UTS_VLC ⚖️ Legal 10,000 ~230,709
UVN-1 📰 News 10,000 ~222,842
Total 20,000 ~453,551

Data Splits

Following UD guidelines:

Split Sentences Percentage
Train 18,282 91.4%
Validation 859 4.3%
Test 859 4.3%
Total 20,000 100%

Usage 

from datasets import load_dataset

# Load all splits
dataset = load_dataset("undertheseanlp/UDD-1")
print(dataset)

# Access specific split
train_data = dataset["train"]
dev_data = dataset["validation"]
test_data = dataset["test"]

# Example sentence
print(train_data[0])

Sentence ID Prefixes

  • s* - Sentences from legal corpus (UTS_VLC)
  • uvn-* - Sentences from news corpus (UVN-1)

Fields

Field Description
sent_id Sentence identifier
text Original sentence text
tokens Word tokens
lemmas Lemmatized forms
upos Universal POS tags
xpos Language-specific POS tags
feats Morphological features
head Head token indices (1-indexed, 0=root)
deprel Dependency relation labels
deps Enhanced dependencies
misc Miscellaneous annotations

Citation 

@misc{underthesea_udd1_2026,
  title = {UDD-1: Universal Dependency Dataset for Vietnamese},
  author = {Underthesea NLP},
  year = {2026},
  publisher = {Hugging Face},
  url = {https://huggingface.co/datasets/undertheseanlp/UDD-1}
}

License

This dataset is licensed under CC BY-SA 4.0.

Contact