failed09/bashkir-multilingual-phrasebooks

Bashkir-Russian Phrasebook Corpus

Edited Bashkir-Russian words, expressions and conversational phrases derived from the original bashkorttele/trilingual-parallel-phrasebooks-bgpu dataset, published by Bashkorttele from phrasebooks of M. Akmulla Bashkir State Pedagogical University.

The default cleaned configuration contains the current deduplicated Bashkir-Russian edition. This release contains Bashkir-Russian pairs.

Configurations

cleaned

• one Bashkir-Russian pair per row
• layout continuations joined and identified OCR and translation errors edited
• exact pair duplicates removed without regard to letter case
• original spelling and the first occurrence retained when merging duplicates
• word forms, punctuation variants and alternative translations remain distinct
• original record IDs and source-book membership retained

reviewed

The edited edition before deduplication, with repeated expressions and within-book order preserved. It includes the same editorial corrections and word / collocation / phrase annotation. This is not the raw OCR extraction. Use it for source-level analysis or alternative deduplication; cleaned remains the recommended default.

Release version, computed row counts and checksums are available in META.json.

Schema

Column	Type	Description
id	string	Retained source record ID
text_ba	string	Edited Bashkir text
text_ru	string	Edited Russian text
entry_type	string	word, collocation or phrase
source_book	string	Source configuration of the retained occurrence
source_ids	list of strings	IDs of all matching edited occurrences before deduplication
source_books	list of strings	All source configurations represented by those occurrences
following_section	string or null	JSON-encoded section heading separated from the retained translation, where present

Source-book codes are ba-ru-alt, ba-ru-ar, ba-ru-kk, ba-ru-sah and ba-ru-zh. These identify the original books, not additional translation columns in this release. source_ids tracks deduplicated occurrences; it does not enumerate every absorbed line from layout reconstruction. In reviewed, source_ids and source_books contain only the current occurrence; in cleaned, they aggregate matching occurrences.

Source and Processing

The Bashkir and Russian columns were taken from all source configurations. Editing was assisted by language models. The process included joining adjacent layout fragments, correcting identified OCR errors and assigning entry types. The edited batches were assembled in book order.

Deduplication compares (text_ba.casefold(), text_ru.casefold()). It does not strip punctuation, remove parenthetical variants or lemmatize words. The spelling of the retained occurrence is not lowercased. Counts and provenance were checked during export, and the saved Parquet was read back and compared with the selected JSON edition.

Limitations

The corpus has been edited to correct OCR errors, restore fragmented phrases and annotate entry types. Some translations and annotations may benefit from further review. Inflected forms and alternative phrasings are preserved.

The train split is a distribution convention; no independent evaluation split or leakage-free benchmark is provided.

License

Distributed under CDLA-Permissive-2.0, as is the original dataset. The original dataset card states that rights to the underlying material were obtained from the rights holders. This release contains derived text data, not scans, illustrations or the original page layout.

Loading

from datasets import load_dataset

dataset = load_dataset("failed09/bashkir-multilingual-phrasebooks", "cleaned")
df = dataset["train"].to_pandas()

# Edited data with repeated expressions retained:
reviewed = load_dataset("failed09/bashkir-multilingual-phrasebooks", "reviewed")

Citation

@dataset{failed09_bashkir_russian_phrasebooks_2026,
  title = {Bashkir-Russian Phrasebook Corpus},
  author = {failed09},
  year = {2026},
  publisher = {Hugging Face},
  url = {https://huggingface.co/datasets/failed09/bashkir-multilingual-phrasebooks},
  note = {Derived from bashkorttele/trilingual-parallel-phrasebooks-bgpu}
}

Open data for the preservation and development of the Bashkir language.