README.md

---
license: cc-by-nc-4.0
task_categories:
- sentence-similarity
language:
- lb
- ltz
size_categories:
- 10K<n<100K
configs:  # Optional. This can be used to pass additional parameters to the dataset loader, such as `data_files`, `data_dir`, and any builder-specific parameters  
- config_name: lb-en
  data_files:
  - split: train
    path: data/v3/lb_en.json
  default: true   # optional

- config_name: lb-fr
  data_files:
  - split: train
    path: data/v3/lb_fr.json
---

# Dataset Card for LuxAlign

## Loading the Dataset
The dataset is currently at **version v3**, which can be loaded as:

```python
from datasets import load_dataset

ds = load_dataset("fredxlpy/LuxAlign", name="lb-en")  # or "lb-fr"
```

If you want to reproduce the results from the paper (`v1`) or use any previous version, you can specify the version folder:

```python
# Load version v1 (as used in the paper)
ds_v1 = load_dataset("fredxlpy/LuxAlign", data_dir="data/v1", data_files={"train": "lb_en.json"})
```

## Dataset Summary
LuxAlign is a parallel dataset featuring Luxembourgish-English and Luxembourgish-French sentence pairs, introduced in [*LuxEmbedder: A Cross-Lingual Approach to Enhanced Luxembourgish Sentence Embeddings* (Philippy et al., 2025)](https://aclanthology.org/2025.coling-main.753/). Designed to align the Luxembourgish embedding space with those of other languages, it enables improved cross-lingual sentence representations for Luxemborgish. This dataset was used to train the Luxembourgish sentence embedding model [**LuxEmbedder**](https://huggingface.co/fredxlpy/LuxEmbedder). The data originates from news articles published by the Luxembourgish news platform [RTL.lu](https://www.rtl.lu).

The sentence pairs in this dataset are not always exact translations but instead reflect high semantic similarity; hence, this dataset may not be suitable for training a machine translation model without caution.

## Dataset Description
- **Repository:** [fredxlpy/LuxEmbedder](https://github.com/fredxlpy/LuxEmbedder)
- **Paper:** [LuxEmbedder: A Cross-Lingual Approach to Enhanced Luxembourgish Sentence Embeddings (Philippy et al., COLING 2025)](https://doi.org/10.48550/arXiv.2412.03331)

## Citation Information
```bibtex
@inproceedings{philippy-etal-2025-luxembedder,
    title = "{L}ux{E}mbedder: A Cross-Lingual Approach to Enhanced {L}uxembourgish Sentence Embeddings",
    author = "Philippy, Fred  and
      Guo, Siwen  and
      Klein, Jacques  and
      Bissyande, Tegawende",
    editor = "Rambow, Owen  and
      Wanner, Leo  and
      Apidianaki, Marianna  and
      Al-Khalifa, Hend  and
      Eugenio, Barbara Di  and
      Schockaert, Steven",
    booktitle = "Proceedings of the 31st International Conference on Computational Linguistics",
    month = jan,
    year = "2025",
    address = "Abu Dhabi, UAE",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2025.coling-main.753/",
    pages = "11369--11379",
}
```

We would like to express our sincere gratitude to RTL Luxembourg for providing the raw seed data that served as the foundation for this research. Those interested in obtaining this data are encouraged to reach out to [RTL Luxembourg](https://www.rtl.lu) or Mr. Tom Weber via [ai@rtl.lu](mailto:ai@rtl.lu).