vpermilp/lid-oss

spellman LID corpus — oss

92,533 rows / 36.9M chars of oss-language text with provenance, built for training and evaluating the spellman language detector. Rows are single-line texts: news articles and library works (paragraph/sentence-group sized) plus Telegram/VK community posts and comments (wild register — emoji, hashtags, code-switching artefacts, folk spellings kept on purpose).

Row length: median 282 chars (p10 53, p90 712).

Configs

config	contents	rows
moderated (default)	news · library · wiki — edited, published sources	57,455
wild	tg-post · tg-comment · vk-post · vk-comment — Telegram and VK community posts and comments	35,078
all	union of both	92,533

from datasets import load_dataset
moderated = load_dataset("vpermilp/lid-oss")            # default config
wild = load_dataset("vpermilp/lid-oss", "wild")

Columns

column	meaning
text	the row (single line, whitespace-flattened)
lang	ISO 639-3 (oss)
source	site store or Telegram channel handle
register	news · library · wiki · tg-post · tg-comment · vk-post · vk-comment
date	publication date when the source exposed one (YYYY-MM-DD)
id	source URL (crawl) or message/post id (Telegram, VK)

Registers

register	rows
news	57,455
tg-post	20,388
vk-post	13,384
tg-comment	1,306

Sources

source	register	rows
rastdzinad.ru-oss	news	55,525
XurZaerin	tg-post	11,690
osetinski	vk-post	7,282
irystontv	tg-post	3,404
rastdzinad	vk-post	2,748
barzafcag	vk-post	2,496
gazetarastdzinad1923	tg-post	2,121
mahduuug	tg-post	1,996
xurzarin-oss	news	1,930
ironzarjyta	tg-post	779
irystontv	tg-comment	759
nogdzau	vk-post	505
mahdug	vk-post	353
mahduuug	tg-comment	342
osetinski	tg-post	294
zaervatkkk	tg-comment	129
zaervatkkk	tg-post	97
osetinski	tg-comment	73
ossetian_language	tg-post	7
ossetian_language	tg-comment	3

How it was built and cleaned

1. Orthographic gate — every row carries letters that never occur in Russian orthography (e.g. Sakha ө ү ҥ ҕ һ), with a per-language marker-density floor; Russian-function-word share gates code-switched rows, which are split into native sentence runs rather than dropped.
2. Structural cleaning — site chrome and per-channel boilerplate (lines repeated across ≥8 posts), self-promo tails, TV schedules, Telegram thread roots (auto-forwarded post copies) removed.
3. Structural repairs — soft hyphens removed and print-layout splits rejoined (hyphenated / lowercase-continued paragraphs merged in page order); site breadcrumbs, TV schedules, weather tables, chord sheets, sponsor tails and letter-spacing artefacts dropped; Latin h/x homoglyphs inside Cyrillic words normalized (399 rows dropped); long rows must carry marker letters in ≥2 tokens (330).
4. Content-free filter — emoji/URL/number-only rows, Latin-dominant or mojibake rows, HTML-entity leftovers dropped (64).
5. Cross-source dedup on a normalized key (casefold; URLs, handles, hashtags, punctuation, digits stripped) — catches site↔channel cross-posts and reposts (2,438 removed).
6. PII masking — phone numbers and e-mails replaced by <phone> / <email>; no author identities were ever stored.
7. Model judge — rows the current spellman model contradicts at ≥0.9 confidence outside the language's close-pair group are dropped (4).

No train/validation/test split is shipped: the spellman pipeline assigns splits content-addressed (crc32) at mix time.

License

The compilation, cleaning and annotations are released under the MIT license. Row texts are redistributed from the public sources listed above (newspapers, libraries, public Telegram channels and VK communities) for language-identification research; each row carries its source and id so upstream attribution is always recoverable.

Built 2026-08-30T12:21:31+00:00.