1
|
1 |
|
import json |
2
|
|
|
|
3
|
1 |
|
from ppp_questionparsing_grammatical import computeTree, simplify, DependenciesTree,\ |
4
|
|
|
QuotationHandler, normalFormProduction, GrammaticalError, NamedEntityMerging, PrepositionMerging |
5
|
1 |
|
import data |
6
|
|
|
|
7
|
1 |
|
from unittest import TestCase |
8
|
|
|
|
9
|
1 |
|
class StandardTripleTests(TestCase): |
10
|
|
|
|
11
|
1 |
View Code Duplication |
def testAndnormalFormProduction(self): |
|
|
|
|
12
|
1 |
|
tree = computeTree(data.give_chief()) |
13
|
1 |
|
NamedEntityMerging(tree).merge() |
14
|
1 |
|
PrepositionMerging(tree).merge() |
15
|
1 |
|
qw = simplify(tree) |
16
|
1 |
|
result = normalFormProduction(tree, qw) |
17
|
1 |
|
self.assertEqual(result, { |
18
|
|
|
"list": [ |
19
|
|
|
{ |
20
|
|
|
"type": "triple", |
21
|
|
|
"subject": { |
22
|
|
|
"type": "resource", |
23
|
|
|
"value": "chief" |
24
|
|
|
}, |
25
|
|
|
"object": { |
26
|
|
|
"type": "missing" |
27
|
|
|
}, |
28
|
|
|
"predicate": { |
29
|
|
|
"list": [ |
30
|
|
|
{ |
31
|
|
|
"type": "resource", |
32
|
|
|
"value": "identity" |
33
|
|
|
} |
34
|
|
|
], |
35
|
|
|
"type": "list" |
36
|
|
|
} |
37
|
|
|
}, |
38
|
|
|
{ |
39
|
|
|
"type": "triple", |
40
|
|
|
"subject": { |
41
|
|
|
"type": "resource", |
42
|
|
|
"value": "prime minister" |
43
|
|
|
}, |
44
|
|
|
"object": { |
45
|
|
|
"type": "missing" |
46
|
|
|
}, |
47
|
|
|
"predicate": { |
48
|
|
|
"list": [ |
49
|
|
|
{ |
50
|
|
|
"type": "resource", |
51
|
|
|
"value": "identity" |
52
|
|
|
} |
53
|
|
|
], |
54
|
|
|
"type": "list" |
55
|
|
|
} |
56
|
|
|
} |
57
|
|
|
], |
58
|
|
|
"type": "intersection" |
59
|
|
|
}) |
60
|
|
|
|
61
|
1 |
View Code Duplication |
def testSuperlativenormalFormProduction(self): |
|
|
|
|
62
|
1 |
|
tree = computeTree(data.give_opera()) |
63
|
1 |
|
NamedEntityMerging(tree).merge() |
64
|
1 |
|
PrepositionMerging(tree).merge() |
65
|
1 |
|
qw = simplify(tree) |
66
|
1 |
|
result = normalFormProduction(tree, qw) |
67
|
1 |
|
self.assertEqual(result, { |
68
|
|
|
"list": [ |
69
|
|
|
{ |
70
|
|
|
"list": { |
71
|
|
|
"predicate": { |
72
|
|
|
"value": "default", |
73
|
|
|
"type": "resource" |
74
|
|
|
}, |
75
|
|
|
"list": { |
76
|
|
|
"value": "Gilbert", |
77
|
|
|
"type": "resource" |
78
|
|
|
}, |
79
|
|
|
"type": "sort" |
80
|
|
|
}, |
81
|
|
|
"index": 0, |
82
|
|
|
"type": "nth" |
83
|
|
|
}, |
84
|
|
|
{ |
85
|
|
|
"list": { |
86
|
|
|
"predicate": { |
87
|
|
|
"value": "default", |
88
|
|
|
"type": "resource" |
89
|
|
|
}, |
90
|
|
|
"list": { |
91
|
|
|
"value": "Sullivan opera", |
92
|
|
|
"type": "resource" |
93
|
|
|
}, |
94
|
|
|
"type": "sort" |
95
|
|
|
}, |
96
|
|
|
"index": 0, |
97
|
|
|
"type": "nth" |
98
|
|
|
} |
99
|
|
|
], |
100
|
|
|
"type": "intersection" |
101
|
|
|
}) |
102
|
|
|
|
103
|
1 |
|
def testnormalFormProduction1(self): |
104
|
1 |
|
tree = computeTree(data.give_president_of_USA()) |
105
|
1 |
|
NamedEntityMerging(tree).merge() |
106
|
1 |
|
PrepositionMerging(tree).merge() |
107
|
1 |
|
qw = simplify(tree) |
108
|
1 |
|
result = normalFormProduction(tree, qw) |
109
|
1 |
|
self.assertEqual(result, { |
110
|
|
|
"object": { |
111
|
|
|
"type": "missing" |
112
|
|
|
}, |
113
|
|
|
"subject": { |
114
|
|
|
"type": "resource", |
115
|
|
|
"value": "United States" |
116
|
|
|
}, |
117
|
|
|
"predicate": { |
118
|
|
|
"type": "resource", |
119
|
|
|
"value": "president" |
120
|
|
|
}, |
121
|
|
|
"type": "triple" |
122
|
|
|
}) |
123
|
|
|
|
124
|
1 |
|
def testnormalFormProduction2(self): |
125
|
1 |
|
handler = QuotationHandler('foo') |
126
|
1 |
|
sentence = 'Who wrote "Lucy in the Sky with Diamonds" and "Let It Be"?' |
127
|
1 |
|
nonAmbiguousSentence = handler.pull(sentence) |
128
|
1 |
|
result=data.give_LSD_LIB() |
129
|
1 |
|
tree=computeTree(result) |
130
|
1 |
|
handler.push(tree) |
131
|
1 |
|
NamedEntityMerging(tree).merge() |
132
|
1 |
|
PrepositionMerging(tree).merge() |
133
|
1 |
|
qw = simplify(tree) |
134
|
1 |
|
result = normalFormProduction(tree, qw) |
135
|
1 |
|
self.assertEqual(result, { |
136
|
|
|
"list": [ |
137
|
|
|
{ |
138
|
|
|
"inverse-predicate": { |
139
|
|
|
"list": [ |
140
|
|
|
{ |
141
|
|
|
"value": "author", |
142
|
|
|
"type": "resource" |
143
|
|
|
}, |
144
|
|
|
{ |
145
|
|
|
"value": "writer", |
146
|
|
|
"type": "resource" |
147
|
|
|
} |
148
|
|
|
], |
149
|
|
|
"type": "list" |
150
|
|
|
}, |
151
|
|
|
"subject": { |
152
|
|
|
"type": "missing" |
153
|
|
|
}, |
154
|
|
|
"object": { |
155
|
|
|
"value": "Lucy in the Sky with Diamonds", |
156
|
|
|
"type": "resource" |
157
|
|
|
}, |
158
|
|
|
"type": "triple", |
159
|
|
|
"predicate": { |
160
|
|
|
"list": [ |
161
|
|
|
{ |
162
|
|
|
"value": "written", |
163
|
|
|
"type": "resource" |
164
|
|
|
}, |
165
|
|
|
{ |
166
|
|
|
"value": "literary works", |
167
|
|
|
"type": "resource" |
168
|
|
|
}, |
169
|
|
|
{ |
170
|
|
|
"value": "bibliography", |
171
|
|
|
"type": "resource" |
172
|
|
|
}, |
173
|
|
|
{ |
174
|
|
|
"value": "notable work", |
175
|
|
|
"type": "resource" |
176
|
|
|
} |
177
|
|
|
], |
178
|
|
|
"type": "list" |
179
|
|
|
} |
180
|
|
|
}, |
181
|
|
|
{ |
182
|
|
|
"inverse-predicate": { |
183
|
|
|
"list": [ |
184
|
|
|
{ |
185
|
|
|
"value": "author", |
186
|
|
|
"type": "resource" |
187
|
|
|
}, |
188
|
|
|
{ |
189
|
|
|
"value": "writer", |
190
|
|
|
"type": "resource" |
191
|
|
|
} |
192
|
|
|
], |
193
|
|
|
"type": "list" |
194
|
|
|
}, |
195
|
|
|
"subject": { |
196
|
|
|
"type": "missing" |
197
|
|
|
}, |
198
|
|
|
"object": { |
199
|
|
|
"value": "Let It Be", |
200
|
|
|
"type": "resource" |
201
|
|
|
}, |
202
|
|
|
"type": "triple", |
203
|
|
|
"predicate": { |
204
|
|
|
"list": [ |
205
|
|
|
{ |
206
|
|
|
"value": "written", |
207
|
|
|
"type": "resource" |
208
|
|
|
}, |
209
|
|
|
{ |
210
|
|
|
"value": "literary works", |
211
|
|
|
"type": "resource" |
212
|
|
|
}, |
213
|
|
|
{ |
214
|
|
|
"value": "bibliography", |
215
|
|
|
"type": "resource" |
216
|
|
|
}, |
217
|
|
|
{ |
218
|
|
|
"value": "notable work", |
219
|
|
|
"type": "resource" |
220
|
|
|
} |
221
|
|
|
], |
222
|
|
|
"type": "list" |
223
|
|
|
} |
224
|
|
|
} |
225
|
|
|
], |
226
|
|
|
"type": "intersection" |
227
|
|
|
}) |
228
|
|
|
|
229
|
1 |
|
def testnormalFormProduction3(self): |
230
|
1 |
|
tree = computeTree(data.give_obama_president_usa()) |
231
|
1 |
|
NamedEntityMerging(tree).merge() |
232
|
1 |
|
PrepositionMerging(tree).merge() |
233
|
1 |
|
qw = simplify(tree) |
234
|
1 |
|
result = normalFormProduction(tree, qw) |
235
|
1 |
|
self.assertEqual(result, { |
236
|
|
|
"type": "intersection", |
237
|
|
|
"list": [ |
238
|
|
|
{ |
239
|
|
|
"predicate": { |
240
|
|
|
"type": "list", |
241
|
|
|
"list": [ |
242
|
|
|
{ |
243
|
|
|
"type": "resource", |
244
|
|
|
"value": "been" |
245
|
|
|
}, |
246
|
|
|
{ |
247
|
|
|
"type": "resource", |
248
|
|
|
"value": "identity" |
249
|
|
|
} |
250
|
|
|
] |
251
|
|
|
}, |
252
|
|
|
"type": "triple", |
253
|
|
|
"subject": { |
254
|
|
|
"type": "resource", |
255
|
|
|
"value": "Obama" |
256
|
|
|
}, |
257
|
|
|
"inverse-predicate": { |
258
|
|
|
"type": "resource", |
259
|
|
|
"value": "identity" |
260
|
|
|
}, |
261
|
|
|
"object": { |
262
|
|
|
"type": "missing" |
263
|
|
|
} |
264
|
|
|
}, |
265
|
|
|
{ |
266
|
|
|
"predicate": { |
267
|
|
|
"type": "list", |
268
|
|
|
"list": [ |
269
|
|
|
{ |
270
|
|
|
"type": "resource", |
271
|
|
|
"value": "been" |
272
|
|
|
}, |
273
|
|
|
{ |
274
|
|
|
"type": "resource", |
275
|
|
|
"value": "identity" |
276
|
|
|
} |
277
|
|
|
] |
278
|
|
|
}, |
279
|
|
|
"type": "triple", |
280
|
|
|
"subject": { |
281
|
|
|
"type": "resource", |
282
|
|
|
"value": "United States president" |
283
|
|
|
}, |
284
|
|
|
"inverse-predicate": { |
285
|
|
|
"type": "resource", |
286
|
|
|
"value": "identity" |
287
|
|
|
}, |
288
|
|
|
"object": { |
289
|
|
|
"type": "missing" |
290
|
|
|
} |
291
|
|
|
} |
292
|
|
|
] |
293
|
|
View Code Duplication |
}) |
|
|
|
|
294
|
|
|
|
295
|
1 |
|
def testnormalFormProductionR8(self): |
296
|
1 |
|
tree = computeTree(data.mistake()) |
297
|
1 |
|
NamedEntityMerging(tree).merge() |
298
|
1 |
|
PrepositionMerging(tree).merge() |
299
|
1 |
|
qw = simplify(tree) |
300
|
1 |
|
result = normalFormProduction(tree, qw) |
301
|
1 |
|
self.assertEqual(result, { |
302
|
|
|
"type": "triple", |
303
|
|
|
"subject": { |
304
|
|
|
"type": "resource", |
305
|
|
|
"value": "mistake" |
306
|
|
|
}, |
307
|
|
|
"object": { |
308
|
|
|
"type": "missing" |
309
|
|
|
}, |
310
|
|
|
"predicate": { |
311
|
|
|
"type": "list", |
312
|
|
|
"list": [ |
313
|
|
|
{ |
314
|
|
|
"type": "resource", |
315
|
|
|
"value": "place" |
316
|
|
|
}, |
317
|
|
|
{ |
318
|
|
|
"type": "resource", |
319
|
|
|
"value": "location" |
320
|
|
|
}, |
321
|
|
|
{ |
322
|
|
|
"type": "resource", |
323
|
|
|
"value": "residence" |
324
|
|
|
}, |
325
|
|
|
{ |
326
|
|
|
"type": "resource", |
327
|
|
|
"value": "country" |
328
|
|
|
}, |
329
|
|
|
{ |
330
|
|
|
"type": "resource", |
331
|
|
|
"value": "city" |
332
|
|
|
}, |
333
|
|
|
{ |
334
|
|
|
"type": "resource", |
335
|
|
|
"value": "town" |
336
|
|
|
}, |
337
|
|
|
{ |
338
|
|
|
"type": "resource", |
339
|
|
|
"value": "state" |
340
|
|
|
}, |
341
|
|
|
{ |
342
|
|
|
"type": "resource", |
343
|
|
|
"value": "locality" |
344
|
|
|
} |
345
|
|
|
] |
346
|
|
|
} |
347
|
|
|
}) |
348
|
|
|
|
349
|
|
|
|
350
|
1 |
|
def testnormalFormProductionSuperl(self): |
351
|
1 |
|
tree = computeTree(data.tanzania()) |
352
|
1 |
|
NamedEntityMerging(tree).merge() |
353
|
1 |
|
PrepositionMerging(tree).merge() |
354
|
1 |
|
qw = simplify(tree) |
355
|
1 |
|
result = normalFormProduction(tree, qw) |
356
|
1 |
|
self.assertEqual(result, { |
357
|
|
|
"list": { |
358
|
|
|
"list": { |
359
|
|
|
"object": { |
360
|
|
|
"type": "missing" |
361
|
|
|
}, |
362
|
|
|
"predicate": { |
363
|
|
|
"value": "mountain", |
364
|
|
|
"type": "resource" |
365
|
|
|
}, |
366
|
|
|
"subject": { |
367
|
|
|
"value": "Tanzania", |
368
|
|
|
"type": "resource" |
369
|
|
|
}, |
370
|
|
|
"type": "triple" |
371
|
|
|
}, |
372
|
|
|
"predicate": { |
373
|
|
|
"value" : "height", |
374
|
|
|
"type" : "resource" |
375
|
|
|
}, |
376
|
|
|
"type": "sort" |
377
|
|
|
}, |
378
|
|
|
"index": -1, |
379
|
|
|
"type": "nth" |
380
|
|
|
}) |
381
|
|
|
|
382
|
1 |
|
def testnormalFormProductionSuperl2(self): |
383
|
1 |
|
tree = computeTree(data.car()) |
384
|
1 |
|
NamedEntityMerging(tree).merge() |
385
|
1 |
|
PrepositionMerging(tree).merge() |
386
|
1 |
|
qw = simplify(tree) |
387
|
1 |
|
result = normalFormProduction(tree, qw) |
388
|
1 |
|
self.assertEqual(result, { |
389
|
|
|
"list": { |
390
|
|
|
"list": { |
391
|
|
|
"subject": { |
392
|
|
|
"value": "world", |
393
|
|
|
"type": "resource" |
394
|
|
|
}, |
395
|
|
|
"predicate": { |
396
|
|
|
"value": "car", |
397
|
|
|
"type": "resource" |
398
|
|
|
}, |
399
|
|
|
"object": { |
400
|
|
|
"type": "missing" |
401
|
|
|
}, |
402
|
|
|
"type": "triple" |
403
|
|
|
}, |
404
|
|
|
"predicate": { |
405
|
|
|
"value" : "cost", |
406
|
|
|
"type" : "resource" |
407
|
|
|
}, |
408
|
|
|
"type": "sort" |
409
|
|
|
}, |
410
|
|
|
"index": -1, |
411
|
|
|
"type": "nth" |
412
|
|
|
}) |
413
|
|
|
|
414
|
1 |
|
def testCop(self): |
415
|
1 |
|
tree = computeTree(data.black()) |
416
|
1 |
|
NamedEntityMerging(tree).merge() |
417
|
1 |
|
PrepositionMerging(tree).merge() |
418
|
1 |
|
self.assertRaises(GrammaticalError, lambda: simplify(tree)) |
419
|
|
|
|
420
|
1 |
|
def testExists(self): |
421
|
1 |
|
tree = computeTree(data.king_england()) |
422
|
1 |
|
NamedEntityMerging(tree).merge() |
423
|
1 |
|
PrepositionMerging(tree).merge() |
424
|
1 |
|
qw = simplify(tree) |
425
|
1 |
|
result = normalFormProduction(tree, qw) |
426
|
1 |
|
self.assertEqual(result, { |
427
|
|
|
"list": { |
428
|
|
|
"predicate": { |
429
|
|
|
"type": "resource", |
430
|
|
|
"value": "king" |
431
|
|
|
}, |
432
|
|
|
"subject": { |
433
|
|
|
"type": "resource", |
434
|
|
|
"value": "England" |
435
|
|
|
}, |
436
|
|
|
"type": "triple", |
437
|
|
|
"object": { |
438
|
|
|
"type": "missing" |
439
|
|
|
} |
440
|
|
|
}, |
441
|
|
|
"type": "exists" |
442
|
|
|
}) |
443
|
|
|
|
444
|
1 |
|
def testSemiQuestionWord1(self): |
445
|
1 |
|
tree = computeTree(data.roald()) |
446
|
1 |
|
NamedEntityMerging(tree).merge() |
447
|
1 |
|
PrepositionMerging(tree).merge() |
448
|
1 |
|
qw = simplify(tree) |
449
|
1 |
|
result = normalFormProduction(tree, qw) |
450
|
1 |
|
self.assertEqual(result, { |
451
|
|
|
"subject": { |
452
|
|
|
"value": "Roald Dahl", |
453
|
|
|
"type": "resource" |
454
|
|
|
}, |
455
|
|
|
"type": "triple", |
456
|
|
|
"predicate": { |
457
|
|
|
"value": "book", |
458
|
|
|
"type": "resource" |
459
|
|
|
}, |
460
|
|
|
"object": { |
461
|
|
|
"type": "missing" |
462
|
|
|
} |
463
|
|
|
}) |
464
|
|
|
|
465
|
1 |
|
def testSemiQuestionWord3(self): |
466
|
1 |
|
tree = computeTree(data.list_president2()) |
467
|
1 |
|
NamedEntityMerging(tree).merge() |
468
|
1 |
|
PrepositionMerging(tree).merge() |
469
|
1 |
|
qw = simplify(tree) |
470
|
1 |
|
result = normalFormProduction(tree, qw) |
471
|
1 |
|
self.assertEqual(result, { |
472
|
|
|
"type": "triple", |
473
|
|
|
"object": { |
474
|
|
|
"type": "missing" |
475
|
|
|
}, |
476
|
|
|
"predicate": { |
477
|
|
|
"type": "resource", |
478
|
|
|
"value": "president" |
479
|
|
|
}, |
480
|
|
|
"subject": { |
481
|
|
|
"type": "resource", |
482
|
|
|
"value": "France" |
483
|
|
|
} |
484
|
|
|
}) |
485
|
|
|
|
486
|
1 |
|
def testSemiQuestionWord4(self): |
487
|
1 |
|
tree = computeTree(data.capital1()) |
488
|
1 |
|
NamedEntityMerging(tree).merge() |
489
|
1 |
|
PrepositionMerging(tree).merge() |
490
|
1 |
|
qw = simplify(tree) |
491
|
1 |
|
result = normalFormProduction(tree, qw) |
492
|
1 |
|
self.assertEqual(result, { |
493
|
|
|
"predicate": { |
494
|
|
|
"type": "resource", |
495
|
|
|
"value": "capital" |
496
|
|
|
}, |
497
|
|
|
"type": "triple", |
498
|
|
|
"subject": { |
499
|
|
|
"type": "resource", |
500
|
|
|
"value": "France" |
501
|
|
|
}, |
502
|
|
|
"object": { |
503
|
|
|
"type": "missing" |
504
|
|
|
} |
505
|
|
|
}) |
506
|
|
|
|
507
|
1 |
|
def testSemiQuestionWord5(self): |
508
|
1 |
|
tree = computeTree(data.capital2()) |
509
|
1 |
|
NamedEntityMerging(tree).merge() |
510
|
1 |
|
PrepositionMerging(tree).merge() |
511
|
1 |
|
qw = simplify(tree) |
512
|
1 |
|
result = normalFormProduction(tree, qw) |
513
|
1 |
|
self.assertEqual(result, { |
514
|
|
|
"predicate": { |
515
|
|
|
"type": "resource", |
516
|
|
|
"value": "capital" |
517
|
|
|
}, |
518
|
|
|
"type": "triple", |
519
|
|
|
"subject": { |
520
|
|
|
"type": "resource", |
521
|
|
|
"value": "France" |
522
|
|
|
}, |
523
|
|
|
"object": { |
524
|
|
|
"type": "missing" |
525
|
|
|
} |
526
|
|
|
}) |
527
|
|
|
|