src/postprocess/simi.py


1
2
3
4
5
6
7
8
9
10
11

#!python

import textdistance
from sys import stdin

for line in stdin.readlines():
    x,y = line.split('\t')
    x = x.casefold()
    y = y.casefold()
    v = textdistance.ratcliff_obershelp.normalized_similarity(x,y) + textdistance.jaro_winkler.normalized_similarity(x,y) + textdistance.cosine.normalized_similarity(x,y)
    print (int(100*(v/3)))