I think something similar about this issue has been raised some time ago, but here we go:
★<x1/>Usa Jingu Shrine
"Usa Jingu Shrine" is in one of my glossaries, but due to the preceding ★<x1/> part is not recognized by CTE. I'm working with the latest build.
You might call it a mantra.
The idea behind it seem sot be that are shall not be any false term matches when a tag separates a word (and the second half is a match), however there seems to be a need for "tag=word separator".
Not only tags, but even in "★Kanda Shrine" CTE is unable to find the Kanda Shrine part that is in the glossary. The Japanese are particularly fond of simbols like ★, ☆ and many others which have no use in other languages, and in fact you find theme almost everywhere.
So, Igor, please find a solution for this.
Please append such symbols into the "Do not match" list in Edit > Preferences > Memory tab. It should solve the above matching issues.
Ah, so easy it was! Thanks.