|
1
|
|
|
# -*- coding: utf-8 -*- |
|
2
|
|
|
|
|
3
|
|
|
# Copyright 2014-2018 by Christopher C. Little. |
|
4
|
|
|
# This file is part of Abydos. |
|
5
|
|
|
# |
|
6
|
|
|
# Abydos is free software: you can redistribute it and/or modify |
|
7
|
|
|
# it under the terms of the GNU General Public License as published by |
|
8
|
|
|
# the Free Software Foundation, either version 3 of the License, or |
|
9
|
|
|
# (at your option) any later version. |
|
10
|
|
|
# |
|
11
|
|
|
# Abydos is distributed in the hope that it will be useful, |
|
12
|
|
|
# but WITHOUT ANY WARRANTY; without even the implied warranty of |
|
13
|
|
|
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
|
14
|
|
|
# GNU General Public License for more details. |
|
15
|
|
|
# |
|
16
|
|
|
# You should have received a copy of the GNU General Public License |
|
17
|
|
|
# along with Abydos. If not, see <http://www.gnu.org/licenses/>. |
|
18
|
|
|
|
|
19
|
|
|
"""abydos.tests.test_distance.token. |
|
20
|
|
|
|
|
21
|
|
|
This module contains unit tests for abydos.distance.token |
|
22
|
|
|
""" |
|
23
|
|
|
|
|
24
|
|
|
from __future__ import division, unicode_literals |
|
25
|
|
|
|
|
26
|
|
|
import math |
|
27
|
|
|
import unittest |
|
28
|
|
|
|
|
29
|
|
|
from abydos.distance.token import bag, dist_bag, dist_cosine, dist_dice, \ |
|
30
|
|
|
dist_jaccard, dist_monge_elkan, dist_overlap, dist_tversky, sim_bag, \ |
|
31
|
|
|
sim_cosine, sim_dice, sim_jaccard, sim_monge_elkan, sim_overlap, \ |
|
32
|
|
|
sim_tanimoto, sim_tversky, tanimoto |
|
33
|
|
|
from abydos.tokenizer.qgram import QGrams |
|
34
|
|
|
|
|
35
|
|
|
from .. import NONQ_FROM, NONQ_TO |
|
36
|
|
|
|
|
37
|
|
|
|
|
38
|
|
|
class TverskyIndexTestCases(unittest.TestCase): |
|
39
|
|
|
"""Test Tversky functions. |
|
40
|
|
|
|
|
41
|
|
|
abydos.distance.sim_tversky & .dist_tversky |
|
42
|
|
|
""" |
|
43
|
|
|
|
|
44
|
|
|
def test_sim_tversky(self): |
|
45
|
|
|
"""Test abydos.distance.sim_tversky.""" |
|
46
|
|
|
self.assertEqual(sim_tversky('', ''), 1) |
|
47
|
|
|
self.assertEqual(sim_tversky('nelson', ''), 0) |
|
48
|
|
|
self.assertEqual(sim_tversky('', 'neilsen'), 0) |
|
49
|
|
|
self.assertAlmostEqual(sim_tversky('nelson', 'neilsen'), 4/11) |
|
50
|
|
|
|
|
51
|
|
|
self.assertEqual(sim_tversky('', '', 2), 1) |
|
52
|
|
|
self.assertEqual(sim_tversky('nelson', '', 2), 0) |
|
53
|
|
|
self.assertEqual(sim_tversky('', 'neilsen', 2), 0) |
|
54
|
|
|
self.assertAlmostEqual(sim_tversky('nelson', 'neilsen', 2), 4/11) |
|
55
|
|
|
|
|
56
|
|
|
# test valid alpha & beta |
|
57
|
|
|
self.assertRaises(ValueError, sim_tversky, 'abcd', 'dcba', 2, -1, -1) |
|
58
|
|
|
self.assertRaises(ValueError, sim_tversky, 'abcd', 'dcba', 2, -1, 0) |
|
59
|
|
|
self.assertRaises(ValueError, sim_tversky, 'abcd', 'dcba', 2, 0, -1) |
|
60
|
|
|
|
|
61
|
|
|
# test empty QGrams |
|
62
|
|
|
self.assertAlmostEqual(sim_tversky('nelson', 'neilsen', 7), 0.0) |
|
63
|
|
|
|
|
64
|
|
|
# test unequal alpha & beta |
|
65
|
|
|
self.assertAlmostEqual(sim_tversky('niall', 'neal', 2, 2, 1), 3/11) |
|
66
|
|
|
self.assertAlmostEqual(sim_tversky('niall', 'neal', 2, 1, 2), 3/10) |
|
67
|
|
|
self.assertAlmostEqual(sim_tversky('niall', 'neal', 2, 2, 2), 3/13) |
|
68
|
|
|
|
|
69
|
|
|
# test bias parameter |
|
70
|
|
|
self.assertAlmostEqual(sim_tversky('niall', 'neal', 2, 1, 1, 0.5), |
|
71
|
|
|
7/11) |
|
72
|
|
|
self.assertAlmostEqual(sim_tversky('niall', 'neal', 2, 2, 1, 0.5), 7/9) |
|
73
|
|
|
self.assertAlmostEqual(sim_tversky('niall', 'neal', 2, 1, 2, 0.5), |
|
74
|
|
|
7/15) |
|
75
|
|
|
self.assertAlmostEqual(sim_tversky('niall', 'neal', 2, 2, 2, 0.5), |
|
76
|
|
|
7/11) |
|
77
|
|
|
|
|
78
|
|
|
# supplied q-gram tests |
|
79
|
|
|
self.assertEqual(sim_tversky(QGrams(''), QGrams('')), 1) |
|
80
|
|
|
self.assertEqual(sim_tversky(QGrams('nelson'), QGrams('')), 0) |
|
81
|
|
|
self.assertEqual(sim_tversky(QGrams(''), QGrams('neilsen')), 0) |
|
82
|
|
|
self.assertAlmostEqual(sim_tversky(QGrams('nelson'), |
|
83
|
|
|
QGrams('neilsen')), 4/11) |
|
84
|
|
|
|
|
85
|
|
|
# non-q-gram tests |
|
86
|
|
|
self.assertEqual(sim_tversky('', '', 0), 1) |
|
87
|
|
|
self.assertEqual(sim_tversky('the quick', '', 0), 0) |
|
88
|
|
|
self.assertEqual(sim_tversky('', 'the quick', 0), 0) |
|
89
|
|
|
self.assertAlmostEqual(sim_tversky(NONQ_FROM, NONQ_TO, 0), 1/3) |
|
90
|
|
|
self.assertAlmostEqual(sim_tversky(NONQ_TO, NONQ_FROM, 0), 1/3) |
|
91
|
|
|
|
|
92
|
|
|
def test_dist_tversky(self): |
|
93
|
|
|
"""Test abydos.distance.dist_tversky.""" |
|
94
|
|
|
self.assertEqual(dist_tversky('', ''), 0) |
|
95
|
|
|
self.assertEqual(dist_tversky('nelson', ''), 1) |
|
96
|
|
|
self.assertEqual(dist_tversky('', 'neilsen'), 1) |
|
97
|
|
|
self.assertAlmostEqual(dist_tversky('nelson', 'neilsen'), 7/11) |
|
98
|
|
|
|
|
99
|
|
|
self.assertEqual(dist_tversky('', '', 2), 0) |
|
100
|
|
|
self.assertEqual(dist_tversky('nelson', '', 2), 1) |
|
101
|
|
|
self.assertEqual(dist_tversky('', 'neilsen', 2), 1) |
|
102
|
|
|
self.assertAlmostEqual(dist_tversky('nelson', 'neilsen', 2), 7/11) |
|
103
|
|
|
|
|
104
|
|
|
# test valid alpha & beta |
|
105
|
|
|
self.assertRaises(ValueError, dist_tversky, 'abcd', 'dcba', 2, -1, -1) |
|
106
|
|
|
self.assertRaises(ValueError, dist_tversky, 'abcd', 'dcba', 2, -1, 0) |
|
107
|
|
|
self.assertRaises(ValueError, dist_tversky, 'abcd', 'dcba', 2, 0, -1) |
|
108
|
|
|
|
|
109
|
|
|
# test empty QGrams |
|
110
|
|
|
self.assertAlmostEqual(dist_tversky('nelson', 'neilsen', 7), 1.0) |
|
111
|
|
|
|
|
112
|
|
|
# test unequal alpha & beta |
|
113
|
|
|
self.assertAlmostEqual(dist_tversky('niall', 'neal', 2, 2, 1), 8/11) |
|
114
|
|
|
self.assertAlmostEqual(dist_tversky('niall', 'neal', 2, 1, 2), 7/10) |
|
115
|
|
|
self.assertAlmostEqual(dist_tversky('niall', 'neal', 2, 2, 2), 10/13) |
|
116
|
|
|
|
|
117
|
|
|
# test bias parameter |
|
118
|
|
|
self.assertAlmostEqual(dist_tversky('niall', 'neal', 2, 1, 1, 0.5), |
|
119
|
|
|
4/11) |
|
120
|
|
|
self.assertAlmostEqual(dist_tversky('niall', 'neal', 2, 2, 1, 0.5), |
|
121
|
|
|
2/9) |
|
122
|
|
|
self.assertAlmostEqual(dist_tversky('niall', 'neal', 2, 1, 2, 0.5), |
|
123
|
|
|
8/15) |
|
124
|
|
|
self.assertAlmostEqual(dist_tversky('niall', 'neal', 2, 2, 2, 0.5), |
|
125
|
|
|
4/11) |
|
126
|
|
|
|
|
127
|
|
|
# supplied q-gram tests |
|
128
|
|
|
self.assertEqual(dist_tversky(QGrams(''), QGrams('')), 0) |
|
129
|
|
|
self.assertEqual(dist_tversky(QGrams('nelson'), QGrams('')), 1) |
|
130
|
|
|
self.assertEqual(dist_tversky(QGrams(''), QGrams('neilsen')), 1) |
|
131
|
|
|
self.assertAlmostEqual(dist_tversky(QGrams('nelson'), |
|
132
|
|
|
QGrams('neilsen')), 7/11) |
|
133
|
|
|
|
|
134
|
|
|
# non-q-gram tests |
|
135
|
|
|
self.assertEqual(dist_tversky('', '', 0), 0) |
|
136
|
|
|
self.assertEqual(dist_tversky('the quick', '', 0), 1) |
|
137
|
|
|
self.assertEqual(dist_tversky('', 'the quick', 0), 1) |
|
138
|
|
|
self.assertAlmostEqual(dist_tversky(NONQ_FROM, NONQ_TO, 0), 2/3) |
|
139
|
|
|
self.assertAlmostEqual(dist_tversky(NONQ_TO, NONQ_FROM, 0), 2/3) |
|
140
|
|
|
|
|
141
|
|
|
|
|
142
|
|
|
class DiceTestCases(unittest.TestCase): |
|
143
|
|
|
"""Test Dice functions. |
|
144
|
|
|
|
|
145
|
|
|
abydos.distance.sim_dice & .dist_dice |
|
146
|
|
|
""" |
|
147
|
|
|
|
|
148
|
|
|
def test_sim_dice(self): |
|
149
|
|
|
"""Test abydos.distance.sim_dice.""" |
|
150
|
|
|
self.assertEqual(sim_dice('', ''), 1) |
|
151
|
|
|
self.assertEqual(sim_dice('nelson', ''), 0) |
|
152
|
|
|
self.assertEqual(sim_dice('', 'neilsen'), 0) |
|
153
|
|
|
self.assertAlmostEqual(sim_dice('nelson', 'neilsen'), 8/15) |
|
154
|
|
|
|
|
155
|
|
|
self.assertEqual(sim_dice('', '', 2), 1) |
|
156
|
|
|
self.assertEqual(sim_dice('nelson', '', 2), 0) |
|
157
|
|
|
self.assertEqual(sim_dice('', 'neilsen', 2), 0) |
|
158
|
|
|
self.assertAlmostEqual(sim_dice('nelson', 'neilsen', 2), 8/15) |
|
159
|
|
|
|
|
160
|
|
|
# supplied q-gram tests |
|
161
|
|
|
self.assertEqual(sim_dice(QGrams(''), QGrams('')), 1) |
|
162
|
|
|
self.assertEqual(sim_dice(QGrams('nelson'), QGrams('')), 0) |
|
163
|
|
|
self.assertEqual(sim_dice(QGrams(''), QGrams('neilsen')), 0) |
|
164
|
|
|
self.assertAlmostEqual(sim_dice(QGrams('nelson'), QGrams('neilsen')), |
|
165
|
|
|
8/15) |
|
166
|
|
|
|
|
167
|
|
|
# non-q-gram tests |
|
168
|
|
|
self.assertEqual(sim_dice('', '', 0), 1) |
|
169
|
|
|
self.assertEqual(sim_dice('the quick', '', 0), 0) |
|
170
|
|
|
self.assertEqual(sim_dice('', 'the quick', 0), 0) |
|
171
|
|
|
self.assertAlmostEqual(sim_dice(NONQ_FROM, NONQ_TO, 0), 1/2) |
|
172
|
|
|
self.assertAlmostEqual(sim_dice(NONQ_TO, NONQ_FROM, 0), 1/2) |
|
173
|
|
|
|
|
174
|
|
|
def test_dist_dice(self): |
|
175
|
|
|
"""Test abydos.distance.dist_dice.""" |
|
176
|
|
|
self.assertEqual(dist_dice('', ''), 0) |
|
177
|
|
|
self.assertEqual(dist_dice('nelson', ''), 1) |
|
178
|
|
|
self.assertEqual(dist_dice('', 'neilsen'), 1) |
|
179
|
|
|
self.assertAlmostEqual(dist_dice('nelson', 'neilsen'), 7/15) |
|
180
|
|
|
|
|
181
|
|
|
self.assertEqual(dist_dice('', '', 2), 0) |
|
182
|
|
|
self.assertEqual(dist_dice('nelson', '', 2), 1) |
|
183
|
|
|
self.assertEqual(dist_dice('', 'neilsen', 2), 1) |
|
184
|
|
|
self.assertAlmostEqual(dist_dice('nelson', 'neilsen', 2), 7/15) |
|
185
|
|
|
|
|
186
|
|
|
# supplied q-gram tests |
|
187
|
|
|
self.assertEqual(dist_dice(QGrams(''), QGrams('')), 0) |
|
188
|
|
|
self.assertEqual(dist_dice(QGrams('nelson'), QGrams('')), 1) |
|
189
|
|
|
self.assertEqual(dist_dice(QGrams(''), QGrams('neilsen')), 1) |
|
190
|
|
|
self.assertAlmostEqual(dist_dice(QGrams('nelson'), QGrams('neilsen')), |
|
191
|
|
|
7/15) |
|
192
|
|
|
|
|
193
|
|
|
# non-q-gram tests |
|
194
|
|
|
self.assertEqual(dist_dice('', '', 0), 0) |
|
195
|
|
|
self.assertEqual(dist_dice('the quick', '', 0), 1) |
|
196
|
|
|
self.assertEqual(dist_dice('', 'the quick', 0), 1) |
|
197
|
|
|
self.assertAlmostEqual(dist_dice(NONQ_FROM, NONQ_TO, 0), 1/2) |
|
198
|
|
|
self.assertAlmostEqual(dist_dice(NONQ_TO, NONQ_FROM, 0), 1/2) |
|
199
|
|
|
|
|
200
|
|
|
|
|
201
|
|
|
class JaccardTestCases(unittest.TestCase): |
|
202
|
|
|
"""Test Jaccard functions. |
|
203
|
|
|
|
|
204
|
|
|
abydos.distance.sim_jaccard & .dist_jaccard |
|
205
|
|
|
""" |
|
206
|
|
|
|
|
207
|
|
|
def test_sim_jaccard(self): |
|
208
|
|
|
"""Test abydos.distance.sim_jaccard.""" |
|
209
|
|
|
self.assertEqual(sim_jaccard('', ''), 1) |
|
210
|
|
|
self.assertEqual(sim_jaccard('nelson', ''), 0) |
|
211
|
|
|
self.assertEqual(sim_jaccard('', 'neilsen'), 0) |
|
212
|
|
|
self.assertAlmostEqual(sim_jaccard('nelson', 'neilsen'), 4/11) |
|
213
|
|
|
|
|
214
|
|
|
self.assertEqual(sim_jaccard('', '', 2), 1) |
|
215
|
|
|
self.assertEqual(sim_jaccard('nelson', '', 2), 0) |
|
216
|
|
|
self.assertEqual(sim_jaccard('', 'neilsen', 2), 0) |
|
217
|
|
|
self.assertAlmostEqual(sim_jaccard('nelson', 'neilsen', 2), 4/11) |
|
218
|
|
|
|
|
219
|
|
|
# supplied q-gram tests |
|
220
|
|
|
self.assertEqual(sim_jaccard(QGrams(''), QGrams('')), 1) |
|
221
|
|
|
self.assertEqual(sim_jaccard(QGrams('nelson'), QGrams('')), 0) |
|
222
|
|
|
self.assertEqual(sim_jaccard(QGrams(''), QGrams('neilsen')), 0) |
|
223
|
|
|
self.assertAlmostEqual(sim_jaccard(QGrams('nelson'), |
|
224
|
|
|
QGrams('neilsen')), 4/11) |
|
225
|
|
|
|
|
226
|
|
|
# non-q-gram tests |
|
227
|
|
|
self.assertEqual(sim_jaccard('', '', 0), 1) |
|
228
|
|
|
self.assertEqual(sim_jaccard('the quick', '', 0), 0) |
|
229
|
|
|
self.assertEqual(sim_jaccard('', 'the quick', 0), 0) |
|
230
|
|
|
self.assertAlmostEqual(sim_jaccard(NONQ_FROM, NONQ_TO, 0), 1/3) |
|
231
|
|
|
self.assertAlmostEqual(sim_jaccard(NONQ_TO, NONQ_FROM, 0), 1/3) |
|
232
|
|
|
|
|
233
|
|
|
def test_dist_jaccard(self): |
|
234
|
|
|
"""Test abydos.distance.dist_jaccard.""" |
|
235
|
|
|
self.assertEqual(dist_jaccard('', ''), 0) |
|
236
|
|
|
self.assertEqual(dist_jaccard('nelson', ''), 1) |
|
237
|
|
|
self.assertEqual(dist_jaccard('', 'neilsen'), 1) |
|
238
|
|
|
self.assertAlmostEqual(dist_jaccard('nelson', 'neilsen'), 7/11) |
|
239
|
|
|
|
|
240
|
|
|
self.assertEqual(dist_jaccard('', '', 2), 0) |
|
241
|
|
|
self.assertEqual(dist_jaccard('nelson', '', 2), 1) |
|
242
|
|
|
self.assertEqual(dist_jaccard('', 'neilsen', 2), 1) |
|
243
|
|
|
self.assertAlmostEqual(dist_jaccard('nelson', 'neilsen', 2), 7/11) |
|
244
|
|
|
|
|
245
|
|
|
# supplied q-gram tests |
|
246
|
|
|
self.assertEqual(dist_jaccard(QGrams(''), QGrams('')), 0) |
|
247
|
|
|
self.assertEqual(dist_jaccard(QGrams('nelson'), QGrams('')), 1) |
|
248
|
|
|
self.assertEqual(dist_jaccard(QGrams(''), QGrams('neilsen')), 1) |
|
249
|
|
|
self.assertAlmostEqual(dist_jaccard(QGrams('nelson'), |
|
250
|
|
|
QGrams('neilsen')), 7/11) |
|
251
|
|
|
|
|
252
|
|
|
# non-q-gram tests |
|
253
|
|
|
self.assertEqual(dist_jaccard('', '', 0), 0) |
|
254
|
|
|
self.assertEqual(dist_jaccard('the quick', '', 0), 1) |
|
255
|
|
|
self.assertEqual(dist_jaccard('', 'the quick', 0), 1) |
|
256
|
|
|
self.assertAlmostEqual(dist_jaccard(NONQ_FROM, NONQ_TO, 0), 2/3) |
|
257
|
|
|
self.assertAlmostEqual(dist_jaccard(NONQ_TO, NONQ_FROM, 0), 2/3) |
|
258
|
|
|
|
|
259
|
|
|
|
|
260
|
|
|
class OverlapTestCases(unittest.TestCase): |
|
261
|
|
|
"""Test overlap functions. |
|
262
|
|
|
|
|
263
|
|
|
abydos.distance.sim_overlap & .dist_overlap |
|
264
|
|
|
""" |
|
265
|
|
|
|
|
266
|
|
|
def test_sim_overlap(self): |
|
267
|
|
|
"""Test abydos.distance.sim_overlap.""" |
|
268
|
|
|
self.assertEqual(sim_overlap('', ''), 1) |
|
269
|
|
|
self.assertEqual(sim_overlap('nelson', ''), 0) |
|
270
|
|
|
self.assertEqual(sim_overlap('', 'neilsen'), 0) |
|
271
|
|
|
self.assertAlmostEqual(sim_overlap('nelson', 'neilsen'), 4/7) |
|
272
|
|
|
|
|
273
|
|
|
self.assertEqual(sim_overlap('', '', 2), 1) |
|
274
|
|
|
self.assertEqual(sim_overlap('nelson', '', 2), 0) |
|
275
|
|
|
self.assertEqual(sim_overlap('', 'neilsen', 2), 0) |
|
276
|
|
|
self.assertAlmostEqual(sim_overlap('nelson', 'neilsen', 2), 4/7) |
|
277
|
|
|
|
|
278
|
|
|
# supplied q-gram tests |
|
279
|
|
|
self.assertEqual(sim_overlap(QGrams(''), QGrams('')), 1) |
|
280
|
|
|
self.assertEqual(sim_overlap(QGrams('nelson'), QGrams('')), 0) |
|
281
|
|
|
self.assertEqual(sim_overlap(QGrams(''), QGrams('neilsen')), 0) |
|
282
|
|
|
self.assertAlmostEqual(sim_overlap(QGrams('nelson'), |
|
283
|
|
|
QGrams('neilsen')), 4/7) |
|
284
|
|
|
|
|
285
|
|
|
# non-q-gram tests |
|
286
|
|
|
self.assertEqual(sim_overlap('', '', 0), 1) |
|
287
|
|
|
self.assertEqual(sim_overlap('the quick', '', 0), 0) |
|
288
|
|
|
self.assertEqual(sim_overlap('', 'the quick', 0), 0) |
|
289
|
|
|
self.assertAlmostEqual(sim_overlap(NONQ_FROM, NONQ_TO, 0), 4/7) |
|
290
|
|
|
self.assertAlmostEqual(sim_overlap(NONQ_TO, NONQ_FROM, 0), 4/7) |
|
291
|
|
|
|
|
292
|
|
|
def test_dist_overlap(self): |
|
293
|
|
|
"""Test abydos.distance.dist_overlap.""" |
|
294
|
|
|
self.assertEqual(dist_overlap('', ''), 0) |
|
295
|
|
|
self.assertEqual(dist_overlap('nelson', ''), 1) |
|
296
|
|
|
self.assertEqual(dist_overlap('', 'neilsen'), 1) |
|
297
|
|
|
self.assertAlmostEqual(dist_overlap('nelson', 'neilsen'), 3/7) |
|
298
|
|
|
|
|
299
|
|
|
self.assertEqual(dist_overlap('', '', 2), 0) |
|
300
|
|
|
self.assertEqual(dist_overlap('nelson', '', 2), 1) |
|
301
|
|
|
self.assertEqual(dist_overlap('', 'neilsen', 2), 1) |
|
302
|
|
|
self.assertAlmostEqual(dist_overlap('nelson', 'neilsen', 2), 3/7) |
|
303
|
|
|
|
|
304
|
|
|
# supplied q-gram tests |
|
305
|
|
|
self.assertEqual(dist_overlap(QGrams(''), QGrams('')), 0) |
|
306
|
|
|
self.assertEqual(dist_overlap(QGrams('nelson'), QGrams('')), 1) |
|
307
|
|
|
self.assertEqual(dist_overlap(QGrams(''), QGrams('neilsen')), 1) |
|
308
|
|
|
self.assertAlmostEqual(dist_overlap(QGrams('nelson'), |
|
309
|
|
|
QGrams('neilsen')), 3/7) |
|
310
|
|
|
|
|
311
|
|
|
# non-q-gram tests |
|
312
|
|
|
self.assertEqual(dist_overlap('', '', 0), 0) |
|
313
|
|
|
self.assertEqual(dist_overlap('the quick', '', 0), 1) |
|
314
|
|
|
self.assertEqual(dist_overlap('', 'the quick', 0), 1) |
|
315
|
|
|
self.assertAlmostEqual(dist_overlap(NONQ_FROM, NONQ_TO, 0), 3/7) |
|
316
|
|
|
self.assertAlmostEqual(dist_overlap(NONQ_TO, NONQ_FROM, 0), 3/7) |
|
317
|
|
|
|
|
318
|
|
|
|
|
319
|
|
|
class TanimotoTestCases(unittest.TestCase): |
|
320
|
|
|
"""Test Tanimoto functions. |
|
321
|
|
|
|
|
322
|
|
|
abydos.distance.sim_tanimoto & .tanimoto |
|
323
|
|
|
""" |
|
324
|
|
|
|
|
325
|
|
|
def test_tanimoto_coeff(self): |
|
326
|
|
|
"""Test abydos.distance.sim_tanimoto.""" |
|
327
|
|
|
self.assertEqual(sim_tanimoto('', ''), 1) |
|
328
|
|
|
self.assertEqual(sim_tanimoto('nelson', ''), 0) |
|
329
|
|
|
self.assertEqual(sim_tanimoto('', 'neilsen'), 0) |
|
330
|
|
|
self.assertAlmostEqual(sim_tanimoto('nelson', 'neilsen'), 4/11) |
|
331
|
|
|
|
|
332
|
|
|
self.assertEqual(sim_tanimoto('', '', 2), 1) |
|
333
|
|
|
self.assertEqual(sim_tanimoto('nelson', '', 2), 0) |
|
334
|
|
|
self.assertEqual(sim_tanimoto('', 'neilsen', 2), 0) |
|
335
|
|
|
self.assertAlmostEqual(sim_tanimoto('nelson', 'neilsen', 2), 4/11) |
|
336
|
|
|
|
|
337
|
|
|
# supplied q-gram tests |
|
338
|
|
|
self.assertEqual(sim_tanimoto(QGrams(''), QGrams('')), 1) |
|
339
|
|
|
self.assertEqual(sim_tanimoto(QGrams('nelson'), QGrams('')), 0) |
|
340
|
|
|
self.assertEqual(sim_tanimoto(QGrams(''), QGrams('neilsen')), 0) |
|
341
|
|
|
self.assertAlmostEqual(sim_tanimoto(QGrams('nelson'), |
|
342
|
|
|
QGrams('neilsen')), 4/11) |
|
343
|
|
|
|
|
344
|
|
|
# non-q-gram tests |
|
345
|
|
|
self.assertEqual(sim_tanimoto('', '', 0), 1) |
|
346
|
|
|
self.assertEqual(sim_tanimoto('the quick', '', 0), 0) |
|
347
|
|
|
self.assertEqual(sim_tanimoto('', 'the quick', 0), 0) |
|
348
|
|
|
self.assertAlmostEqual(sim_tanimoto(NONQ_FROM, NONQ_TO, 0), 1/3) |
|
349
|
|
|
self.assertAlmostEqual(sim_tanimoto(NONQ_TO, NONQ_FROM, 0), 1/3) |
|
350
|
|
|
|
|
351
|
|
|
def test_tanimoto(self): |
|
352
|
|
|
"""Test abydos.distance.tanimoto.""" |
|
353
|
|
|
self.assertEqual(tanimoto('', ''), 0) |
|
354
|
|
|
self.assertEqual(tanimoto('nelson', ''), float('-inf')) |
|
355
|
|
|
self.assertEqual(tanimoto('', 'neilsen'), float('-inf')) |
|
356
|
|
|
self.assertAlmostEqual(tanimoto('nelson', 'neilsen'), |
|
357
|
|
|
math.log(4/11, 2)) |
|
358
|
|
|
|
|
359
|
|
|
self.assertEqual(tanimoto('', '', 2), 0) |
|
360
|
|
|
self.assertEqual(tanimoto('nelson', '', 2), float('-inf')) |
|
361
|
|
|
self.assertEqual(tanimoto('', 'neilsen', 2), float('-inf')) |
|
362
|
|
|
self.assertAlmostEqual(tanimoto('nelson', 'neilsen', 2), |
|
363
|
|
|
math.log(4/11, 2)) |
|
364
|
|
|
|
|
365
|
|
|
# supplied q-gram tests |
|
366
|
|
|
self.assertEqual(tanimoto(QGrams(''), QGrams('')), 0) |
|
367
|
|
|
self.assertEqual(tanimoto(QGrams('nelson'), QGrams('')), float('-inf')) |
|
368
|
|
|
self.assertEqual(tanimoto(QGrams(''), QGrams('neilsen')), |
|
369
|
|
|
float('-inf')) |
|
370
|
|
|
self.assertAlmostEqual(tanimoto(QGrams('nelson'), QGrams('neilsen')), |
|
371
|
|
|
math.log(4/11, 2)) |
|
372
|
|
|
|
|
373
|
|
|
# non-q-gram tests |
|
374
|
|
|
self.assertEqual(tanimoto('', '', 0), 0) |
|
375
|
|
|
self.assertEqual(tanimoto('the quick', '', 0), float('-inf')) |
|
376
|
|
|
self.assertEqual(tanimoto('', 'the quick', 0), float('-inf')) |
|
377
|
|
|
self.assertAlmostEqual(tanimoto(NONQ_FROM, NONQ_TO, 0), |
|
378
|
|
|
math.log(1/3, 2)) |
|
379
|
|
|
self.assertAlmostEqual(tanimoto(NONQ_TO, NONQ_FROM, 0), |
|
380
|
|
|
math.log(1/3, 2)) |
|
381
|
|
|
|
|
382
|
|
|
|
|
383
|
|
|
class CosineSimilarityTestCases(unittest.TestCase): |
|
384
|
|
|
"""Test cosine similarity functions. |
|
385
|
|
|
|
|
386
|
|
|
abydos.distance.sim_cosine & .dist_cosine |
|
387
|
|
|
""" |
|
388
|
|
|
|
|
389
|
|
|
def test_sim_cosine(self): |
|
390
|
|
|
"""Test abydos.distance.sim_cosine.""" |
|
391
|
|
|
self.assertEqual(sim_cosine('', ''), 1) |
|
392
|
|
|
self.assertEqual(sim_cosine('nelson', ''), 0) |
|
393
|
|
|
self.assertEqual(sim_cosine('', 'neilsen'), 0) |
|
394
|
|
|
self.assertAlmostEqual(sim_cosine('nelson', 'neilsen'), |
|
395
|
|
|
4/math.sqrt(7*8)) |
|
396
|
|
|
|
|
397
|
|
|
self.assertEqual(sim_cosine('', '', 2), 1) |
|
398
|
|
|
self.assertEqual(sim_cosine('nelson', '', 2), 0) |
|
399
|
|
|
self.assertEqual(sim_cosine('', 'neilsen', 2), 0) |
|
400
|
|
|
self.assertAlmostEqual(sim_cosine('nelson', 'neilsen', 2), |
|
401
|
|
|
4/math.sqrt(7*8)) |
|
402
|
|
|
|
|
403
|
|
|
# supplied q-gram tests |
|
404
|
|
|
self.assertEqual(sim_cosine(QGrams(''), QGrams('')), 1) |
|
405
|
|
|
self.assertEqual(sim_cosine(QGrams('nelson'), QGrams('')), 0) |
|
406
|
|
|
self.assertEqual(sim_cosine(QGrams(''), QGrams('neilsen')), 0) |
|
407
|
|
|
self.assertAlmostEqual(sim_cosine(QGrams('nelson'), QGrams('neilsen')), |
|
408
|
|
|
4/math.sqrt(7*8)) |
|
409
|
|
|
|
|
410
|
|
|
# non-q-gram tests |
|
411
|
|
|
self.assertEqual(sim_cosine('', '', 0), 1) |
|
412
|
|
|
self.assertEqual(sim_cosine('the quick', '', 0), 0) |
|
413
|
|
|
self.assertEqual(sim_cosine('', 'the quick', 0), 0) |
|
414
|
|
|
self.assertAlmostEqual(sim_cosine(NONQ_FROM, NONQ_TO, 0), |
|
415
|
|
|
4/math.sqrt(9*7)) |
|
416
|
|
|
self.assertAlmostEqual(sim_cosine(NONQ_TO, NONQ_FROM, 0), |
|
417
|
|
|
4/math.sqrt(9*7)) |
|
418
|
|
|
|
|
419
|
|
|
def test_dist_cosine(self): |
|
420
|
|
|
"""Test abydos.distance.dist_cosine.""" |
|
421
|
|
|
self.assertEqual(dist_cosine('', ''), 0) |
|
422
|
|
|
self.assertEqual(dist_cosine('nelson', ''), 1) |
|
423
|
|
|
self.assertEqual(dist_cosine('', 'neilsen'), 1) |
|
424
|
|
|
self.assertAlmostEqual(dist_cosine('nelson', 'neilsen'), |
|
425
|
|
|
1-(4/math.sqrt(7*8))) |
|
426
|
|
|
|
|
427
|
|
|
self.assertEqual(dist_cosine('', '', 2), 0) |
|
428
|
|
|
self.assertEqual(dist_cosine('nelson', '', 2), 1) |
|
429
|
|
|
self.assertEqual(dist_cosine('', 'neilsen', 2), 1) |
|
430
|
|
|
self.assertAlmostEqual(dist_cosine('nelson', 'neilsen', 2), |
|
431
|
|
|
1-(4/math.sqrt(7*8))) |
|
432
|
|
|
|
|
433
|
|
|
# supplied q-gram tests |
|
434
|
|
|
self.assertEqual(dist_cosine(QGrams(''), QGrams('')), 0) |
|
435
|
|
|
self.assertEqual(dist_cosine(QGrams('nelson'), QGrams('')), 1) |
|
436
|
|
|
self.assertEqual(dist_cosine(QGrams(''), QGrams('neilsen')), 1) |
|
437
|
|
|
self.assertAlmostEqual(dist_cosine(QGrams('nelson'), |
|
438
|
|
|
QGrams('neilsen')), |
|
439
|
|
|
1-(4/math.sqrt(7*8))) |
|
440
|
|
|
|
|
441
|
|
|
# non-q-gram tests |
|
442
|
|
|
self.assertEqual(dist_cosine('', '', 0), 0) |
|
443
|
|
|
self.assertEqual(dist_cosine('the quick', '', 0), 1) |
|
444
|
|
|
self.assertEqual(dist_cosine('', 'the quick', 0), 1) |
|
445
|
|
|
self.assertAlmostEqual(dist_cosine(NONQ_FROM, NONQ_TO, 0), |
|
446
|
|
|
1-4/math.sqrt(9*7)) |
|
447
|
|
|
self.assertAlmostEqual(dist_cosine(NONQ_TO, NONQ_FROM, 0), |
|
448
|
|
|
1-4/math.sqrt(9*7)) |
|
449
|
|
|
|
|
450
|
|
|
|
|
451
|
|
|
class MongeElkanTestCases(unittest.TestCase): |
|
452
|
|
|
"""Test Monge-Elkan functions. |
|
453
|
|
|
|
|
454
|
|
|
abydos.distance.sim_monge_elkan & .dist_monge_elkan |
|
455
|
|
|
""" |
|
456
|
|
|
|
|
457
|
|
|
def test_sim_monge_elkan(self): |
|
458
|
|
|
"""Test abydos.distance.sim_monge_elkan.""" |
|
459
|
|
|
self.assertEqual(sim_monge_elkan('', ''), 1) |
|
460
|
|
|
self.assertEqual(sim_monge_elkan('', 'a'), 0) |
|
461
|
|
|
self.assertEqual(sim_monge_elkan('a', 'a'), 1) |
|
462
|
|
|
|
|
463
|
|
|
self.assertEqual(sim_monge_elkan('Niall', 'Neal'), 3/4) |
|
464
|
|
|
self.assertEqual(sim_monge_elkan('Niall', 'Njall'), 5/6) |
|
465
|
|
|
self.assertEqual(sim_monge_elkan('Niall', 'Niel'), 3/4) |
|
466
|
|
|
self.assertEqual(sim_monge_elkan('Niall', 'Nigel'), 3/4) |
|
467
|
|
|
|
|
468
|
|
|
self.assertEqual(sim_monge_elkan('Niall', 'Neal', symmetric=True), |
|
469
|
|
|
31/40) |
|
470
|
|
|
self.assertEqual(sim_monge_elkan('Niall', 'Njall', symmetric=True), |
|
471
|
|
|
5/6) |
|
472
|
|
|
self.assertEqual(sim_monge_elkan('Niall', 'Niel', symmetric=True), |
|
473
|
|
|
31/40) |
|
474
|
|
|
self.assertAlmostEqual(sim_monge_elkan('Niall', 'Nigel', |
|
475
|
|
|
symmetric=True), 17/24) |
|
476
|
|
|
|
|
477
|
|
|
def test_dist_monge_elkan(self): |
|
478
|
|
|
"""Test abydos.distance.dist_monge_elkan.""" |
|
479
|
|
|
self.assertEqual(dist_monge_elkan('', ''), 0) |
|
480
|
|
|
self.assertEqual(dist_monge_elkan('', 'a'), 1) |
|
481
|
|
|
|
|
482
|
|
|
self.assertEqual(dist_monge_elkan('Niall', 'Neal'), 1/4) |
|
483
|
|
|
self.assertAlmostEqual(dist_monge_elkan('Niall', 'Njall'), 1/6) |
|
484
|
|
|
self.assertEqual(dist_monge_elkan('Niall', 'Niel'), 1/4) |
|
485
|
|
|
self.assertEqual(dist_monge_elkan('Niall', 'Nigel'), 1/4) |
|
486
|
|
|
|
|
487
|
|
|
self.assertAlmostEqual(dist_monge_elkan('Niall', 'Neal', |
|
488
|
|
|
symmetric=True), 9/40) |
|
489
|
|
|
self.assertAlmostEqual(dist_monge_elkan('Niall', 'Njall', |
|
490
|
|
|
symmetric=True), 1/6) |
|
491
|
|
|
self.assertAlmostEqual(dist_monge_elkan('Niall', 'Niel', |
|
492
|
|
|
symmetric=True), 9/40) |
|
493
|
|
|
self.assertAlmostEqual(dist_monge_elkan('Niall', 'Nigel', |
|
494
|
|
|
symmetric=True), 7/24) |
|
495
|
|
|
|
|
496
|
|
|
|
|
497
|
|
|
class BagTestCases(unittest.TestCase): |
|
498
|
|
|
"""Test bag similarity functions. |
|
499
|
|
|
|
|
500
|
|
|
abydos.distance.bag, .sim_bag & .dist_bag |
|
501
|
|
|
""" |
|
502
|
|
|
|
|
503
|
|
|
def test_bag(self): |
|
504
|
|
|
"""Test abydos.distance.bag.""" |
|
505
|
|
|
self.assertEqual(bag('', ''), 0) |
|
506
|
|
|
self.assertEqual(bag('nelson', ''), 6) |
|
507
|
|
|
self.assertEqual(bag('', 'neilsen'), 7) |
|
508
|
|
|
self.assertEqual(bag('ab', 'a'), 1) |
|
509
|
|
|
self.assertEqual(bag('ab', 'c'), 2) |
|
510
|
|
|
self.assertEqual(bag('nelson', 'neilsen'), 2) |
|
511
|
|
|
self.assertEqual(bag('neilsen', 'nelson'), 2) |
|
512
|
|
|
self.assertEqual(bag('niall', 'neal'), 2) |
|
513
|
|
|
self.assertEqual(bag('aluminum', 'Catalan'), 5) |
|
514
|
|
|
self.assertEqual(bag('abcdefg', 'hijklm'), 7) |
|
515
|
|
|
self.assertEqual(bag('abcdefg', 'hijklmno'), 8) |
|
516
|
|
|
|
|
517
|
|
|
def test_sim_bag(self): |
|
518
|
|
|
"""Test abydos.distance.sim_bag.""" |
|
519
|
|
|
self.assertEqual(sim_bag('', ''), 1) |
|
520
|
|
|
self.assertEqual(sim_bag('nelson', ''), 0) |
|
521
|
|
|
self.assertEqual(sim_bag('', 'neilsen'), 0) |
|
522
|
|
|
self.assertEqual(sim_bag('ab', 'a'), 0.5) |
|
523
|
|
|
self.assertEqual(sim_bag('ab', 'c'), 0) |
|
524
|
|
|
self.assertAlmostEqual(sim_bag('nelson', 'neilsen'), 5/7) |
|
525
|
|
|
self.assertAlmostEqual(sim_bag('neilsen', 'nelson'), 5/7) |
|
526
|
|
|
self.assertAlmostEqual(sim_bag('niall', 'neal'), 3/5) |
|
527
|
|
|
self.assertAlmostEqual(sim_bag('aluminum', 'Catalan'), 3/8) |
|
528
|
|
|
self.assertEqual(sim_bag('abcdefg', 'hijklm'), 0) |
|
529
|
|
|
self.assertEqual(sim_bag('abcdefg', 'hijklmno'), 0) |
|
530
|
|
|
|
|
531
|
|
|
def test_dist_bag(self): |
|
532
|
|
|
"""Test abydos.distance.dist_bag.""" |
|
533
|
|
|
self.assertEqual(dist_bag('', ''), 0) |
|
534
|
|
|
self.assertEqual(dist_bag('nelson', ''), 1) |
|
535
|
|
|
self.assertEqual(dist_bag('', 'neilsen'), 1) |
|
536
|
|
|
self.assertEqual(dist_bag('ab', 'a'), 0.5) |
|
537
|
|
|
self.assertEqual(dist_bag('ab', 'c'), 1) |
|
538
|
|
|
self.assertAlmostEqual(dist_bag('nelson', 'neilsen'), 2/7) |
|
539
|
|
|
self.assertAlmostEqual(dist_bag('neilsen', 'nelson'), 2/7) |
|
540
|
|
|
self.assertAlmostEqual(dist_bag('niall', 'neal'), 2/5) |
|
541
|
|
|
self.assertAlmostEqual(dist_bag('aluminum', 'Catalan'), 5/8) |
|
542
|
|
|
self.assertEqual(dist_bag('abcdefg', 'hijklm'), 1) |
|
543
|
|
|
self.assertEqual(dist_bag('abcdefg', 'hijklmno'), 1) |
|
544
|
|
|
|
|
545
|
|
|
|
|
546
|
|
|
if __name__ == '__main__': |
|
547
|
|
|
unittest.main() |
|
548
|
|
|
|