Correct spelling of the word is not showing up in the suggestion box

[patchling]

The correct spelling of the word is not showing up in the suggestion box. Maybe grammar or definition problems?

procedure - the word that I am looking for, got this by entering the word into the DuckDuckGo search bar.

procuder - the letters that I used.

This is the sentence that I used it in: If a person wants something from me the procuder is simple: Approach me and make an offer.

With a little copy & paste this sentence does just fine. If a person wants something from me the procedure is simple: Approach me and make an offer.

Oh, and please keep up the good work with LT, I need it so very much.

[danielnaber]

These words are too far apart, you can check that here: http://www.convertforfree.com/levenshtein-distance-calculator/ - the distance is 3, but to keep LT fast, we stop searching at distance 2 once we have found words.

[dpelle]

@danielnaber wrote:

These words are too far apart, you can check that here: http://www.convertforfree.com/levenshtein-distance-calculator/ - the distance is 3, but to keep LT fast, we stop searching at distance 2 once we have found words.

Does the Levenshtein distance threshold depend on the word length? A threshold of 2 might be too large for a short word of say fewer than 4 letters. But a distance of 2 or even 3 may be quite acceptable for a long word of say more than 9 letters. Intuitively, I would use something like that

word    Levenshtein
length  threshold
======  ===========
<=4     1
5       2
6       2
7       2
8       2
>=9     3

Although, it'd have to be more complex as errors with combination of letter close to each other should could as smaller penalties than other kinds of errors. For example:

• double letter errors should not count too much i.e. "usefull" and "useful" should be close in distance.
• diacritics should not give a big penalty in distance i.e. café and cafe are close to each other
• f and ph which sound the same in several language should be close to either other
• common transliteration can be taken into account such as: ß = ss for German, cx = ĉ for Esperanto, ij = ÿ for Dutch, etc.
• digraphs can be taken into account i.e. ff = ff, fi = fi
• Unicode normalisation should be taken into account
• etc.

[danielnaber]

Does the Levenshtein distance threshold depend on the word length?

Yes, in a simple way:

https://github.com/languagetool-org/languagetool/blob/master/languagetool-core/src/main/java/org/languagetool/rules/spelling/morfologik/MorfologikSpellerRule.java#L185