KOI8-R (RFC 1489) is an 8-bit character encoding derived from the KOI-8 encoding by the programmer Andrei Chernov in 1993 and designed to cover Russian, which uses the Russian subset of a Cyrillic script. KOI-8, on its turn, is an 8-bit extension of the KOI-7 encoding, which inherited a phonetic correspondence of Russian and Latin letters from the MTK-2 teletype code. As a result, Russian Cyrillic letters in KOI8-R are in pseudo-Latin alphabetical order rather than the normal Cyrillic one like in ISO 8859-5. Although this may seem unnatural, this has the useful effect that if the 8th bit is st
via Wikipedia infobox
KOI8-R (RFC 1489) is an 8-bit character encoding derived from the KOI-8 encoding by the programmer Andrei Chernov in 1993 and designed to cover Russian, which uses the Russian subset of a Cyrillic script. KOI-8, on its turn, is an 8-bit extension of the KOI-7 encoding, which inherited a phonetic correspondence of Russian and Latin letters from the MTK-2 teletype code. As a result, Russian Cyrillic letters in KOI8-R are in pseudo-Latin alphabetical order rather than the normal Cyrillic one like in ISO 8859-5. Although this may seem unnatural, this has the useful effect that if the 8th bit is stripped, the text remains partially readable in any ASCII-based encoding (including KOI8-R itself) as a case-reversed transliteration. For example, "Код для обмена и обработки информации" (the Russian meaning of the "KOI" acronym) becomes kOD DLQ OBMENA I OBRABOTKI INFORMACII.
KOI-8 stands for 8-bitnyy kod dlya obmena i obrabotki informatsii () which means "8-Bit Code for Information Interchange". In Microsoft Windows, KOI8-R is assigned the code page number 20866. In IBM, KOI8-R is assigned code page 878. KOI8-R also happens to cover Bulgarian.
Discovered by embedding cosine similarity (sentence-transformers MiniLM, 384-dim).