Check how much a String sounds like another one in Java

Question

I'd like to know if there is any class in Java able to check, using its own criteria, how much a String is equal to another one. Example :

• William Shakespeare / William Shakespeare : might be 100%
• William Shakespe**a**re / William Shakespe**e**re : might have above 90%
• William Shakespeare / Shakespeare, William : might have above 70% (just examples)

Answer 1

I see two main candidates:

• The Soundex encoding, implemented by Apache Commons. However, note that it's mainly meant for single, relatively short words. It won't find a similarity in your third example. Additionally, it really only works for English words.
• The Levenshtein distance (Again implemented at Apache Commons). This is language agnostic, but similarity for switched parts as in your third example will be relatively low (more like 40%). Modifications like the Damerau–Levenshtein distance may yield better results.

Answer 2

Generally, there is the levenshtein algorithm, which just outputs how many insert/update/delete operations you would have to perform (characterwise) in order to transform one string into another. Apache's StringUtils class has an implementation.