README.md

---
license: cc-by-sa-4.0
language:
- ba
- ru
pretty_name: Bashkir-Russian Wikipedia Parallel Corpus
task_categories:
- translation
configs:
- config_name: cleaned
  data_files:
  - split: train
    path: cleaned/data.parquet
- config_name: precleaned
  data_files:
  - split: train
    path: precleaned/data.parquet
- config_name: scored
  data_files:
  - split: train
    path: scored/data.parquet
- config_name: filtered
  data_files:
  - split: train
    path: filtered/data.parquet
tags:
- bashkir
- russian
- wikipedia
- cyrillic
- low-resource
- parallel-corpus
- sentence-level
- bitext-mining
---

# Bashkir-Russian Wikipedia Parallel Corpus

Sentence-level Bashkir-Russian parallel data extracted from the Bashkir and
Russian Wikipedia dumps dated `2026-08-01`.

This repository provides four configurations. The default `cleaned`
configuration is recommended for machine translation training. The
`precleaned` configuration is an earlier, less filtered extraction provided
for alternative preprocessing and research. The `scored` configuration adds
LASER-based quality scores to the `cleaned` pairs. The `filtered`
configuration applies a conservative two-encoder filter on top of `scored`.

## Configurations

### cleaned

- 72,007 sentence pairs
- 34,566 linked article pairs
- URLs, MediaWiki markup, card fragments and malformed punctuation removed
- duplicate and identical pairs removed
- sentence indices retained

Fields:

- `ba_title`: Bashkir Wikipedia article title
- `ru_title`: Russian Wikipedia article title
- `section`: `lead` or `facts`
- `ba`: Bashkir sentence
- `ru`: Russian sentence
- `ba_sentence_index`: sentence index in the Bashkir article
- `ru_sentence_index`: sentence index in the Russian article

### precleaned

- 91,949 sentence pairs
- earlier extraction with lighter filtering
- contains older Wiki markup, links and alignment noise

Fields:

- `ba_title`, `ru_title`, `section`, `ba`, `ru`

### scored

- 72,007 sentence pairs — **identical rows to `cleaned`**, nothing added or removed
- quality scores computed with Meta LASER sentence embeddings:
  - `laser_cos`: cosine similarity between LASER embeddings of the pair
  - `laser_margin`: margin normalized by same-language kNN density
  - `laser_margin_xling`: margin normalized by **cross-lingual** kNN density
    (Artetxe & Schwenk, 2018) — the recommended LASER filter score

Encoders: LASER3 `bak_Cyrl` (NLLB) for Bashkir, LASER2 for Russian,
1024-dim L2-normalized embeddings, k=4 neighbours
(`laser_encoders`, facebookresearch/LASER).

### filtered

- **69,706 sentence pairs** (−3.20% from `scored`)
- conservative two-encoder filter: a pair is removed only when **both**
  scorers place it in their bottom 5%:
  `laser_margin_xling < 0.876 AND labse_margin_xling < 0.723`
- second encoder: **LaBSE** (Google, 481M parameters, fp16) — independent
  architecture and training data, so its errors are largely uncorrelated
  with LASER's
- all score columns retained: `laser_cos`, `laser_margin`,
  `laser_margin_xling`, `labse_cos`, `labse_margin_xling`
- removed pairs are dominated by template mismatch sentences (river-tributary
  stubs with different facts) and cross-article topic matches

Score distributions over the full 72,007 pairs:

| score | p01 | p05 | p25 | p50 | p75 | p95 | p99 |
|---|---|---|---|---|---|---|---|
| `laser_cos` | 0.597 | 0.705 | 0.828 | 0.863 | 0.887 | 0.914 | 0.928 |
| `laser_margin_xling` | 0.760 | 0.876 | 0.988 | 1.033 | 1.063 | 1.102 | 1.129 |
| `labse_cos` | 0.206 | 0.408 | 0.698 | 0.789 | 0.859 | 0.932 | 0.963 |
| `labse_margin_xling` | 0.373 | 0.723 | 1.035 | 1.132 | 1.202 | 1.295 | 1.358 |

## Source and Processing

The data was extracted from corresponding Bashkir and Russian Wikipedia
articles using the `bawiki-20260801` and `ruwiki-20260801` dumps.

The cleaned configuration uses article-local sentence indices, numeric
anchors for factual pairs, length-ratio checks, language checks, duplicate
removal and service-fragment filtering. The `scored` and `filtered`
configurations add multilingual sentence-embedding scoring on top of the
same alignment without modifying any text.

## License

The source text is derived from Wikipedia and is distributed under the
[Creative Commons Attribution-ShareAlike 4.0 International license](https://creativecommons.org/licenses/by-sa/4.0/).

Please preserve Wikipedia attribution when redistributing or publishing
derivative datasets.

## Loading

```python
from datasets import load_dataset

cleaned = load_dataset("failed09/bashkir-wikipedia-parallel", "cleaned")
scored = load_dataset("failed09/bashkir-wikipedia-parallel", "scored")
filtered = load_dataset("failed09/bashkir-wikipedia-parallel", "filtered")

# strict MT-training subset with your own threshold
df = scored["train"].to_pandas()
strict = df[(df.laser_margin_xling >= 0.90) & (df.labse_margin_xling >= 1.0)]
```

## Citation

```bibtex
@dataset{failed09_bashkir_russian_wikipedia_parallel_2026,
  title = {Bashkir-Russian Wikipedia Parallel Corpus},
  author = {failed09},
  year = {2026},
  publisher = {Hugging Face},
  url = {https://huggingface.co/datasets/failed09/bashkir-wikipedia-parallel}
}
```