Re^3: Finding Nearly Identical Sets

[LanX]

Again, for me it is elegant.

It's fast, easy to implement and its matching the requirements.

Well unless you have more in mind than you told us.

Better solutions (like a more straight away normalization) would require your codepoints to have a guaranteed minimal distance.

This could certainly be achieved, but well regarding the generality of your question, it'll be far fetched to suppose such a structure.