def calculate(base, term):
base = base.lower()
term = term.lower()
edit_distance = editDistance.calculate(base, term)
jaccard_distance = jaccardDistance.calculate(base, term)
lcss_distance = longestCommonSubstringDistance.calculate(base, term)
jaccard_metric = jaccard_distance
lcss_metric = lcss_distance / (len(base)
if len(base) > len(term) else len(term))
edit_distance_metric = edit_distance / len(base)
if jaccard_metric == 1:
metric = edit_distance
elif jaccard_metric == 0:
metric = edit_distance_metric
else:
metric = jaccard_metric * edit_distance
if lcss_metric > 0:
metric /= lcss_metric
else:
metric *= (len(base) if len(base) < len(term) else len(term))
return metric
def calculate(base,term): base = base.lower() term = term.lower() edit_distance = editDistance.calculate(base,term) jaccard_distance = jaccardDistance.calculate(base,term) lcss_distance = longestCommonSubstringDistance.calculate(base,term) jaccard_metric = jaccard_distance lcss_metric = lcss_distance/(len(base) if len(base) > len(term) else len(term)) edit_distance_metric = edit_distance/len(base) if jaccard_metric == 1: metric = edit_distance elif jaccard_metric == 0: metric = edit_distance_metric else: metric = jaccard_metric * edit_distance if lcss_metric > 0: metric /= lcss_metric else: metric *= (len(base) if len(base) < len(term) else len(term)) return metric
def test_edit_distance_of_identical_words_case_senseitive(self):
editDistance = calculate("hi", "Hi")
self.assertEqual(editDistance, 1)
def test_edit_distance_of_identical_words(self):
editDistance = calculate("hi", "hi")
self.assertEqual(editDistance, 0)
def test_edit_distance_of_words_with_similar_letters(self):
editDistance = calculate("act", "catch")
self.assertEqual(editDistance, 3)
def test_edit_distance_of_anagrams(self):
editDistance = calculate("tame", "mate")
self.assertEqual(editDistance, 2)
def test_edit_distance_of_subset_words(self):
editDistance = calculate("cat", "catch")
self.assertEqual(editDistance, 2)
def test_edit_distance_of_similar_words(self):
editDistance = calculate("cat", "bat")
self.assertEqual(editDistance, 1)
def test_edit_distance_of_identical_words_case_senseitive(self):
editDistance = calculate("hi","Hi")
self.assertEqual(editDistance, 1)
def test_edit_distance_of_identical_words(self):
editDistance = calculate("hi","hi")
self.assertEqual(editDistance, 0)
def test_edit_distance_of_words_with_similar_letters(self):
editDistance = calculate("act","catch")
self.assertEqual(editDistance, 3)
def test_edit_distance_of_anagrams(self):
editDistance = calculate("tame","mate")
self.assertEqual(editDistance, 2)
def test_edit_distance_of_subset_words(self):
editDistance = calculate("cat","catch")
self.assertEqual(editDistance, 2)
def test_edit_distance_of_similar_words(self):
editDistance = calculate("cat","bat")
self.assertEqual(editDistance, 1)