def test_one_clutter_after(self):
paper_ind = random.randint(0, len(paper_titles) - 1)
junk_ind = random.randint(0, len(junk) - 1)
first = paper_titles[paper_ind]
second = paper_titles[paper_ind] + junk[junk_ind]
common = set(first).intersection(set(second))
self.assertTrue(arrays_contain_same_reference(first, second, common))
# First paper has junk rather than second
first = paper_titles[paper_ind] + junk[junk_ind]
second = paper_titles[paper_ind]
common = set(first).intersection(set(second))
self.assertTrue(arrays_contain_same_reference(first, second, common))
def test_one_clutter_before(self):
paper_ind = random.randint(0, len(paper_titles) - 1)
junk_ind = random.randint(0, len(junk) - 1)
first = paper_titles[paper_ind]
second = junk[junk_ind] + paper_titles[paper_ind]
common = set(first).intersection(set(second))
self.assertTrue(arrays_contain_same_reference(first, second, common))
# check it works if they are the other way round
first = junk[junk_ind] + paper_titles[paper_ind]
second = paper_titles[paper_ind]
common = set(first).intersection(set(second))
self.assertTrue(arrays_contain_same_reference(first, second, common))
def test_noclutter_similar(self):
similar_ind = random.randint(0, len(paper_title_similar_pairs) - 1)
first = paper_title_similar_pairs[similar_ind][0]
second = paper_title_similar_pairs[similar_ind][1]
common = set(first).intersection(set(second))
self.assertFalse(arrays_contain_same_reference(first, second, common))
def test_both_clutter_after_repeated_different(self):
# Artificially construct something so that the overlap is in the junk
paper = ["this", "is", "the", "paper", "title"]
junk_first = ["this", "here", "junk", "words"]
junk_second = ["yet", "more", "stuff", "junk"]
first = paper + junk_first
second = paper + junk_second
common = set(first).intersection(set(second))
self.assertTrue(arrays_contain_same_reference(first, second, common))
def test_noclutter_cutoff(self):
# What do we want to be the case here? If we remove the last word in one
# of the titles then it's sort of like the other one has junk at the
# end?
first = paper_titles[0]
second = paper_titles[1][:-1]
common = set(first).intersection(set(second))
# Go with a more inclusive assumption for now...
self.assertTrue(arrays_contain_same_reference(first, second, common))
def test_both_sameauthor_difftitle_before(self):
first = paper_title_similar_pairs[0][0]
second = paper_title_similar_pairs[0][1]
authors = ["john", "baker", "jill", "smith", "bea", "zisserman"]
first = authors + first
second = authors + second
common = set(first).intersection(set(second))
self.assertTrue(arrays_contain_same_reference(first, second, common))
def test_both_clutter_after(self):
paper_ind = random.randint(0, len(paper_titles) - 1)
junk_first_ind = random.randint(0, len(junk) - 1)
junk_second_ind = random.randint(0, len(junk) - 1)
while junk_first_ind == junk_second_ind:
junk_second_ind = random.randint(0, len(junk) - 1)
first = paper_titles[paper_ind] + junk[junk_first_ind]
second = paper_titles[paper_ind] + junk[junk_second_ind]
common = set(first).intersection(set(second))
self.assertTrue(arrays_contain_same_reference(first, second, common))
def test_one_clutter_after_repeatword(self):
# use clutter which contains a word which exists in the paper title
paper = [
'whats', 'point', 'semantic', 'segmentation', 'with', 'point',
'supervision'
]
junk = [
'3rd', 'workshop', 'semantic', 'perception', 'mapping',
'exploration', 'spme'
]
first = paper
second = paper + junk
common = set(first).intersection(set(second))
self.assertTrue(arrays_contain_same_reference(first, second, common))
first = paper + junk
second = paper
common = set(first).intersection(set(second))
self.assertTrue(arrays_contain_same_reference(first, second, common))
def test_one_clutter_before_repeatword(self):
paper = [
'deeplab', 'semantic', 'image', 'segmentation', 'with', 'deep',
'convolutional', 'nets', 'atrous', 'convolution', 'fully',
'connected', 'crfs'
]
junk = [
'navab', 'hornegger', 'wells', 'frangi', 'eds', 'medical', 'image',
'computing', 'computer-assisted', 'intervention', 'miccai', '2015'
]
first = paper
second = junk + paper
common = set(first).intersection(set(second))
self.assertTrue(arrays_contain_same_reference(first, second, common))
def test_one_clutter_before_after_repeatword_after(self):
paper = [
'deeplab', 'semantic', 'image', 'segmentation', 'with', 'deep',
'convolutional', 'nets', 'atrous', 'convolution', 'fully',
'connected', 'crfs'
]
junk_before = [
'torch7', 'matlab-like', 'environment', 'machine', 'learning',
'biglearn', 'nips', 'workshop', 'dai', 'sun', '2015'
]
junk_after = [
'3rd', 'workshop', 'semantic', 'perception', 'mapping',
'exploration', 'spme'
]
first = paper
second = junk_before + paper + junk_after
common = set(first).intersection(set(second))
self.assertTrue(arrays_contain_same_reference(first, second, common))
def test_one_clutter_before_after_repeatword_before(self):
paper = [
'deeplab', 'semantic', 'image', 'segmentation', 'with', 'deep',
'convolutional', 'nets', 'atrous', 'convolution', 'fully',
'connected', 'crfs'
]
junk_before = [
'navab', 'hornegger', 'wells', 'frangi', 'eds', 'medical', 'image',
'computing', 'computer-assisted', 'intervention', 'miccai', '2015'
]
junk_after = [
'volume', 'jmlr', 'proceedings', 'pages', '195-206', 'jmlrorg',
'2012', 'karen', 'simonyan', 'andrew', 'zisserman'
]
first = paper
second = junk_before + paper + junk_after
common = set(first).intersection(set(second))
self.assertTrue(arrays_contain_same_reference(first, second, common))
def test_noclutter_invalid(self):
first = paper_titles[0]
second = paper_titles[1]
common = set(first).intersection(set(second))
self.assertFalse(arrays_contain_same_reference(first, second, common))
def test_noclutter(self):
for ind, title in enumerate(paper_titles):
common = set(title).intersection(set(title))
self.assertTrue(arrays_contain_same_reference(
title, title, common))