Comparing strings finds an incorrect match

[dimiterbak]

Hi There,

Thanks for sharing this library!

I am running the bellow test and expect to find no match.
However, it returns a match:

import numpy as np
import unittest

from names_matcher.algorithm import NamesMatcher

class TestCreateIdentityMatcher(unittest.TestCase):

    def test_compare_different_identities(self):

        names_1 = [["V", "v"]]
        names_2 = [["L", "o"]]

        assignments = NamesMatcher()(names_1,
                                     names_2)

        self.assertEqual(-1,
                             assignments[0][0])
        self.assertEqual(1,
                             assignments[1][0])

[vmarkovtsev]

Hi @dimiterbak, your code returns the following for me:

(array([0], dtype=int32), array([0.]))

As you see, the confidence of the match is 0, which is the minimum confidence possible.
I always return the matches, and it depends on the domain problem to choose the perfect threshold for the confidence. It depends on what's more important, Precision vs. Recall, etc., etc.

[dimiterbak]

Thank you!

It looks like I misunderstood by thinking you return the distance but it was the confidence.