search

seatgeek / fuzzywuzzy

Fuzzy String Matching in Python
http://chairnerd.seatgeek.com/fuzzywuzzy-fuzzy-string-matching-in-python/
GNU General Public License v2.0
9.2k stars 878 forks source link

Wired behavior of partial_ratio #313

Open sillybun opened 3 years ago

sillybun commented 3 years ago 
In [47]: fuzzywuzzy.fuzz.partial_ratio("red", "random")                        
Out[47]: 33

In [48]: fuzzywuzzy.fuzz.partial_ratio("rod", "random")                        
Out[48]: 33

In [49]: fuzzywuzzy.fuzz.partial_ratio("prod", "random")                       
Out[49]: 25

In [50]: fuzzywuzzy.fuzz.partial_ratio("pred", "random")                       
Out[50]: 50

why "pred" is more similar to "random" than "prod"?

maxbachmann commented 3 years ago

This is a known issue in python-Levenshtein: https://github.com/seatgeek/fuzzywuzzy/issues/79 In your case for the comparision of

"prod" <-> "random"

the following alignment is used:

"prod" <-> "ndom"

which has a similarity of 25. However the optimal alignment would be:

"prod" <-> "rand"

which has a similarity of 50. In FuzzyWuzzy you will get the correct result when the slower difflib based implementation is used:

>>> from fuzzywuzzy import fuzz
>>> from difflib import SequenceMatcher
>>> fuzz.SequenceMatcher = SequenceMatcher
>>> fuzzywuzzy.fuzz.partial_ratio("prod", "random") 
50