🎨 Cross-Field Highlighter addon - spotlight word in text [support thread]

Anki Add-ons
10

Hi @Xirai , thanks for mentioning these cases!

  1. Sometimes other parts of the sentence get highlighted too.

The target word ζ–°[あら]た is broke down (tokenized) by punctuation marks (square brackets in this case) into tokens:

  • ζ–°[あら]た
  • ζ–°
  • あら
  • た

Then each token is highlighted in the text individually.
There is no special logic for furigana.
Unfortunately, in this case, I can’t see an ability to distinguish the 2nd た to skip it. Language processing is a complex task where false-positive and false-negative matches sometimes happen. No magic :slightly_smiling_face:

  1. It doesn’t work right when the word in the sentence use a different form.

It requires using morphological analysis (word dictionaries) to support word forms for Japanese language. Current addon version relies on regular expressions only.

For English, word forms are handled by replacing 2 last characters (for words longer 3 symbols). E.g. study β†’ stu* which matches study, studies, studying, studied, etc. But it doesn’t work for Japanese.

Maybe I’ll add morphological analysis in future versions, but in v1 I can provide only regular expressions.

Despite I can’t fix these 2 cases, please, proceed reporting any anomalies in highlighting or user interface. They allows me to fix issues I can’t mention with my dataset.

1 Like