An implementation of a "soft Jaccard" set similarity measure

gistfile1.py

>>> import jellyfish
>>> from soft_jaccard import soft_jaccard
>>> c1 = set(['CL Isbell','C. L. Isbell'])
>>> c2 = set(['C Isbell','C Isbell, Jr.'])
>>> soft_jaccard(c1, c2, jellyfish.jaro_winkler)
0.75848950260673509

soft_jaccard.py

def soft_jaccard(a, b, sim_func):
    """
    Return a measure of two sets' similarity, based on the similarity of
    their elements.

    Arguments:
    a - a set or otherwise uniqified list of strings
    b - a set or otherwise uniqified list
    sim_func - a function that takes two str args and returns a measure of
    their similarity, a float in [0,1]
    """
    intersection_length = sum(sum(sim_func(i, j) for j in b)/float(len(b)) for i in a)
    return float(intersection_length)/(len(a) + len(b) - intersection_length)