Being aware of some hanjas' phonetic changes

[dahlia]

Some hanjas like 金/讀/畵 can be pronounced in different ways. The current behavior can produce incorrect results in some cases e.g.:

• Input: 金日成綜合大學은 平壤에 있는 朝鮮民主主義人民共和國의 國立大學이다.
• Expected output: 김일성종합대학은 평양에 있는 조선민주주의인민공화국의 국립대학이다.
• Actual output: 금일성종합대학은 평양에 있는 조선민주주의인민공화국의 국립대학이다.

See also the following table:

Hanja	Word 1	Word 2
金	金剛經 (금강경)	金浦國際空港 (김포국제공항)
讀	讀書 (독서)	句讀點 (구두점)
畵	畵龍點睛 (화룡점정)	企畵 (기획)

[suminb]

Thanks for your report. I'm unable to investigate this issue at the moment, but I'll try to re-visit this sometime this week.

[suminb]

Sorry for the late response. It's almost been a year 😆

I looked into this briefly, and it looks like there is no easy way to deal with this issue other than making a huge rule table. Or maybe I'm missing something... If anyone could suggest a solution for this, it would be much appreciated.

[chaaklau]

A huge rule table would be the easiest solution. Since you are using a mapping file, if you have a list of phrases that do not use the most common reading, you can place those phrases on top of your file.

For example, the hanja 金 has two readings 금 and 김. 김 is everywhere and you can't possibly list out all names with the reading 김. If you have a list of all '金' words, e.g. 대금 (代金), 금고 (金庫), 금요일 (金曜日) etc., put them on the top of your list. If there is no matched phrases, then use the default pronunciation. The table will grow 10 times bigger but it should not affect run-time too much.