What's the difference between the distance and similarity functions?

[stijnfrishert]

What's the difference between functions such as levenshtein_distance(), levenshtein_similarity() and the normalized variants?

In Levenshtein.hpp there's two big comments.

• The first comment block explains the Levenshtein distance just fine
• levenshtein_similarity() has no comment
• levenshtein_normalized_distance() has no comment
• The second comment block purports to explain the normalized_levenshtein(), but is placed in front of levenshtein_normalized_similarity()?

I'm guessing similarity is just the opposite of distance (so a similarity of max distance means a full match), but that's guess work on my side. Maybe the comments could use a refresher. :)

[maxbachmann]

I never really got around to making the C++ documentation nice 😅

Your interpretation is correct about what these are:

• distance -> what you would regularly call Levenshtein distance, which is a score between 0 and max, where max for the uniform Levenshtein distance is max(len1, len2)
• similarity -> the inversed version, so a score between max and 0 where max is the optimum
• normalized_distance / normalized_similarity -> normalized version of the versions above to a range from 0.0-1.0 or 1.0-0.0 respectively

[stijnfrishert]

Makes a lot of sense, and thank you for clearing that up :)

Aside from that, very happy with the library!