BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//pretalx.coscup.org//coscup-2026//speaker//FKUXQS
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-2026-7VSVUT@pretalx.coscup.org
DTSTART;TZID=CST:20260808T100000
DTEND;TZID=CST:20260808T103000
DESCRIPTION:大型語言模型產生的繁體中文，常混入中國用語
 、過度簡化的漢語、錯誤標點與不精準術語，這些輸出
 會在文件、教學、開放原始碼專案與技術文章中逐漸污
 染書寫習慣。zhtw-mcp 是個透過 Model Context Protocol (MCP) 整
 合到 AI 助理的繁體中文書寫檢查器，可依中華民國教育
 部標點、國字標準字體與兩岸詞彙差異，檢查並修正 AI 
 產生的中文。\n\n本議程介紹為什麼 AI 中文輸出容易被不
 精準來源混雜影響，如何將中文潤稿轉化為可測試、可
 重現、可融入進開發流程的工程議題，並展示 zhtw-mcp 如
 何自動改善台灣繁體中文書寫材料的品質和文化傳承。\n
 \nLLMs often generate Traditional Chinese that looks fluent but mixes zh-C
 N vocabulary\, Simplified-Chinese phrasing\, inconsistent punctuation\, an
 d unstable technical terminology. zhtw-mcp turns this problem into an engi
 neering workflow: a Model Context Protocol server that checks\, fixes\, an
 d gates zh-TW text using Taiwan-specific language rules.\n\nThis talk pres
 ents the design of zhtw-mcp\, the challenges of handling ambiguous termino
 logy and false friends\, and how linguistic linting can be integrated into
  AI-assisted writing\, documentation\, teaching material\, and open-source
  development workflows.
DTSTAMP:20260713T132525Z
LOCATION:TR409-2
SUMMARY:用 zhtw-mcp 搭配 AI 寫出更像台灣人的中文 | Write Taiwa
 nese Mandarin: zhtw-mcp as a Linguistic Linter for Traditional Chinese - 
 黃敬群
URL:https://pretalx.coscup.org/coscup-2026/talk/7VSVUT/
END:VEVENT
END:VCALENDAR