jc.py

# chl, 2010-05-18
# jaccard coefficient experiment, inspired by @datajunkie

from __future__ import with_statement, division

import collections

S = collections.defaultdict(set)
T = collections.defaultdict(set) # inverse

for a, b in (line.strip().split("\t") for line in open("input")):
    S[a].add(b)
    T[b].add(a)

def jc(a, b):
    return len(S[a] & S[b]) / len(S[a] | S[b])

def jcs(a):
    return ((x, jc(a, x)) for x in reduce(set.union, (T[x] for x in S[a]), set()))

with open("output", "w") as f:
    for a in S.keys():
        for b, v in jcs(a):
            f.write("%s\t%s\t%s\n" % (a, b, v))