BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//pretalx.coscup.org//coscup-2026//talk//TK3BKR
BEGIN:VTIMEZONE
TZID:CST
BEGIN:STANDARD
DTSTART:20000101T000000
RRULE:FREQ=YEARLY;BYMONTH=1
TZNAME:CST
TZOFFSETFROM:+0800
TZOFFSETTO:+0800
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
UID:pretalx-coscup-2026-TK3BKR@pretalx.coscup.org
DTSTART;TZID=CST:20260809T130000
DTEND;TZID=CST:20260809T133000
DESCRIPTION:Yuino is an OSS Kana-Kanji conversion system that integrates LL
 Ms with a dictionary-based approach.\n\nThe LLM-based Kana-Kanji conversio
 n systems have been an actively developing topic in the OSS community in r
 ecent years. That representative example is azooKey. It achieves high-qual
 ity conversion accelerated by an LLM. On the other hand\, it has a clearly
  weak point.\nThis weak point is 'How to handle new words from context'. T
 he existing method doesn’t do not use dictionaries tied to the LLM. They
  use a tokenizer before LLM inference\, and their behavior is entirely dep
 endent on the training data. One of the example of new words is "Reiwa". "
 Reiwa" is an era name in Japan. It was created in 2018.\nThe tokenizer spl
 its dictionary words into multiple tokens. It loses the original meaning a
 nd part-of-speech (POS) information. As a result\, these cases negatively 
 impact the inference process for Kana-Kanji conversion.\n\nTo solve this i
 ssue\, I propose an LLM extension method. This method bypasses the tokeniz
 er and modified the LLM to accept inputs while preserving both the word's 
 form and its POS information. After this modification\, we can input feedi
 ng word data with attached POS tags directly into the LLM. When we use the
  new words in input\, the system can derive conversion results that accura
 tely reflect the intended usage. Yuino is the world first implementation o
 f my proposed method.\nAttendees will learn about my proposed method deepl
 y. The implementation of Yuino is a Japanese-specific implementation. On t
 he other hand\, the method is possible to reuse other languages in an ideo
 gram user. Because my proposed method is common sense.
DTSTAMP:20260713T142816Z
LOCATION:TR410
SUMMARY:Yuino: OSS kana-kanji conversion system that combines LLM and a dic
 tionary-based approach - Masahiko Hashimoto
URL:https://pretalx.coscup.org/coscup-2026/talk/TK3BKR/
END:VEVENT
END:VCALENDAR