Auto-correction word distance

As I mentioned in another thread, I am seeing a problem with auto-correction with accented letters, like razon having razones suggested instead of razón.
I started this topic because I noticed another problem with the word distance algorithm. I have someone in my contacts list with last name Oka and have the option to use contacts as suggestions. When I enter lower case "oka", Oka is the second suggestion even though only casing differences are clearly almost 0 distance but the word gets replace with ola in spanish or okay in english. I wanted to ask about this before opening a bug. Is this a bug? am I missing something? it seems to me that two words with exactly the same spelling but only case differences should be the first suggestion.

I was going to make a similar suggestion.

As for Korean, it might be better if the double-consonant (or double-vowel for compact typeset) error can be suggested earlier than others.
For example, when I enter "잇다", the words "잇단" "잇자" and "익다" are suggested, when "있다" does not appear.
But I think "있다" is closer than "익다" or "잇자". (Also, 있다 is a very frequent word.)

In addition, it would be better if mistake between adjacent keys can be suggested earlier.
I always make mistake when I enter right-most keys.

I was wrong.
It turns out that no words with double-consonants are suggested. (Both syllable-initial and syllable-final)
Maybe problem with dictionary conversion?

It turns out that no words with double-consonants are suggested. (Both syllable-initial and syllable-final)
Maybe problem with dictionary conversion?

No this time this is not a dictionary conversion problem It's because double-consonants and the corresponding consonants are not considered as related letters. I could change the dictionary algorithm if that's really what you want