Datasets:
Update README.md
Browse files
README.md
CHANGED
|
@@ -123,15 +123,7 @@ LANG_NAME_UK = {
|
|
| 123 |
|
| 124 |
**Source Data**
|
| 125 |
|
| 126 |
-
-
|
| 127 |
-
- EUR-Lex (EU acts in multiple languages).
|
| 128 |
-
- Original TMX compilation: ELRC-SHARE “EU acts in Ukrainian” (3,056,791 TUs).
|
| 129 |
-
|
| 130 |
-
**Processing**
|
| 131 |
-
|
| 132 |
-
1. Converted TMX (X–UK) to directional pairs **UK → X**.
|
| 133 |
-
2. Normalized whitespace and removed empty segments.
|
| 134 |
-
3. Preserved document boundaries where available via simple `doc_id` (optional field).
|
| 135 |
|
| 136 |
**Intended Uses**
|
| 137 |
|
|
@@ -139,23 +131,6 @@ LANG_NAME_UK = {
|
|
| 139 |
- Domain adaptation for legal/administrative translation.
|
| 140 |
- Research on multilingual legal translation and terminology.
|
| 141 |
|
| 142 |
-
**Out-of-scope**
|
| 143 |
-
|
| 144 |
-
- Use in safety-critical applications without additional validation.
|
| 145 |
-
- Any use that violates the upstream license terms.
|
| 146 |
-
|
| 147 |
-
## Considerations
|
| 148 |
-
|
| 149 |
-
**Quality and Bias**
|
| 150 |
-
|
| 151 |
-
- Legal texts are formal and domain-specific; expect formulaic style and boilerplate.
|
| 152 |
-
- Some segments may contain references, citations, or enumerations that require specialized tokenization.
|
| 153 |
-
|
| 154 |
-
**Recommendations**
|
| 155 |
-
|
| 156 |
-
- Deduplicate at the segment level if mixing with other legal corpora.
|
| 157 |
-
- Use document context (where available) for paragraph-level MT.
|
| 158 |
-
|
| 159 |
## How to Use
|
| 160 |
|
| 161 |
```python
|
|
@@ -169,14 +144,18 @@ print(pl[0])
|
|
| 169 |
```
|
| 170 |
|
| 171 |
## Citation
|
| 172 |
-
|
| 173 |
-
|
| 174 |
-
|
| 175 |
-
|
| 176 |
-
-
|
| 177 |
-
|
| 178 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 179 |
## License
|
|
|
|
| 180 |
|
| 181 |
-
-
|
| 182 |
-
- **Attribution:** Please credit ELRC-SHARE “EU acts in Ukrainian”, the Parliament of Ukraine portal, and EUR-Lex.
|
|
|
|
| 123 |
|
| 124 |
**Source Data**
|
| 125 |
|
| 126 |
+
- Original TMX compilation: ELRC-SHARE (3,056,791 TUs).
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 127 |
|
| 128 |
**Intended Uses**
|
| 129 |
|
|
|
|
| 131 |
- Domain adaptation for legal/administrative translation.
|
| 132 |
- Research on multilingual legal translation and terminology.
|
| 133 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 134 |
## How to Use
|
| 135 |
|
| 136 |
```python
|
|
|
|
| 144 |
```
|
| 145 |
|
| 146 |
## Citation
|
| 147 |
+
- ELRC-SHARE (CC-BY-SA-4.0).
|
| 148 |
+
|
| 149 |
+
- **BibTeX**
|
| 150 |
+
```bibtex
|
| 151 |
+
@dataset{EU-acts-langs-to-uk,
|
| 152 |
+
title = {EU Acts: Ukrainian → EU Languages },
|
| 153 |
+
author = {lapa-llm},
|
| 154 |
+
year = {2025},
|
| 155 |
+
url = {https://huggingface.co/datasets/lapa-llm/EU-acts-uk-to-langs}
|
| 156 |
+
}
|
| 157 |
+
```
|
| 158 |
## License
|
| 159 |
+
CC-BY-SA-4.0
|
| 160 |
|
| 161 |
+
This dataset is part of the "Lapa" - Ukrainian LLM initiative to advance natural language processing for the Ukrainian language.
|
|
|