forked from recski/tuw-nlp
-
Notifications
You must be signed in to change notification settings - Fork 0
/
lexicon.py
620 lines (549 loc) · 27.1 KB
/
lexicon.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
import os
from tuw_nlp.text.patterns.misc import CHAR_REPLACEMENTS
class IRTGRuleLexicon:
def __init__(self):
self.get_props_from_file("propositions.txt")
self.npos = ("NOUN", "ADJ", "PROPN", "ADV")
self.get_mod_edges()
self.get_term_fnc()
self.get_binary_fnc()
def get_mod_edges(self):
return NotImplementedError
def get_dependency_rules(self, pos, dep, cpos):
"""this is kept even simpler: all three have to be listed for every
lexicon entry.
TODO: allow lookup based on lemma and/or clemma (that's 3 options)"""
return self.bin_fnc.get((pos, dep, cpos)) or [self.default_binary_rule]
def get_terminal_rules(self, word, pos, **kwargs):
raise NotImplementedError
def get_props_from_file(self, fn):
self.prop = set()
with open(os.path.join(os.path.dirname(os.path.abspath(__file__)), fn)) as PF:
for line in PF:
w = line.strip()
for a, b in CHAR_REPLACEMENTS.items():
w = w.replace(a, b)
self.prop.add(w)
def get_binary_fnc(self):
raise NotImplementedError
def get_term_fnc(self):
raise NotImplementedError
def get_lexical_terminal(self, word):
raise NotImplementedError
def get_default_terminal(self, word):
raise NotImplementedError
class FSLexicon(IRTGRuleLexicon):
def __init__(self):
super(FSLexicon, self).__init__()
self.default_binary_rule = "unify(?1, ?2)"
def get_binary_fnc(self):
self.bin_fnc = {
# Errichtung ist untersagt
("ADJ", "NSUBJ", "NOUN"): ["unify(emb_prop(?1), ?2)"],
("VERB", "NSUBJ_PASS", "NOUN"): ["unify(emb_obj(?1), ?2)"],
("VERB", "NSUBJ", "NOUN"): ["unify(emb_subj(?1), ?2)"],
("VERB", "CONJ", "VERB"): ["unify(emb_coord1(?1), emb_coord2(?2))"],
("NOUN", "CASE", "ADP"): ["emb_prop(unify(?1, emb_mod(?2)))"],
}
self.bin_fnc.update(
{edge: ["emb_prop(unify(emb_mod(?1), ?2))"] for edge in self.mod_edges}
)
# coordination
self.bin_fnc.update(
{
(pos1, "CONJ", pos2): ["unify(emb_coord1(?1), emb_coord2(?2))"]
for pos1 in self.npos
for pos2 in self.npos
}
)
def get_term_fnc(self):
self.term_fnc = {
"nicht": ['"[neg: 1]"'],
"kein": ['"[neg: 1]"'],
"duerfen": ['"[per: 1]"'],
"muessen": ['"[obl: 1]"'],
"zulaessig": ['"[per: 1]"'],
"untersagen": ['"[for: 1]"'],
"unzulaessig": ['"[for: 1]"'],
}
def get_lexical_terminal(self, word):
return f'"[prop: {word}]"'
def get_default_terminal(self, word):
return '"[]"'
class BaseLexicon(IRTGRuleLexicon):
def __init__(self):
super(BaseLexicon, self).__init__()
# if a dependency is not handled, the dependent is ignored
self.default_binary_rule = "?2"
def get_binary_fnc(self):
raise NotImplementedError
def get_term_fnc(self):
raise NotImplementedError
def get_terminal_rules(self, word, pos, xpos, i="unknown"):
"""returns a list of interpretations associated with the
word or word, pos pair. Word-only match takes precedence to discourage
using it as an elsewhere condition: if POS matters then
the word should be listed with all possible POS-tags"""
if xpos == "VVIZU":
return [f'"({word}<root> / {word} :0 (OBL / OBL))"']
return self.term_fnc.get(word, self.term_fnc.get((word, pos))) or [
self.get_default_terminal(word, i)
]
def get_lexical_terminal(self, word):
return self.get_default_terminal(word)
def get_default_terminal(self, word, i="unknown"):
return f'"({word}<root> / {word}_{i})"'
class ENLexicon(BaseLexicon):
def __init__(self):
super(ENLexicon, self).__init__()
def get_mod_edges(self):
self.mod_edges = {
("ADJ", "ADVMOD", "ADV"),
("ADJ", "ADVMOD", "PART"),
("ADJ", "ADVMOD", "ADJ"),
("PROPN", "ADVMOD", "ADJ"),
("PROPN", "ADVMOD", "ADV"),
("NOUN", "ADVMOD", "ADV"),
("NOUN", "ADVMOD", "NUM"),
("NOUN", "ADVMOD", "PRON"),
("NUM", "ADVMOD", "ADV"),
("VERB", "ADVMOD", "ADJ"),
("VERB", "ADVMOD", "ADV"),
("VERB", "ADVMOD", "PART"),
("NOUN", "ADVMOD", "PART"),
("PRON", "ADVMOD", "PART"),
("ADJ", "OBL_NPMOD", "NOUN"),
("ADV", "OBL_NPMOD", "NOUN"),
("VERB", "OBL_NPMOD", "NOUN"),
("NOUN", "NMOD", "NOUN"),
("NOUN", "NMOD", "PROPN"),
("PROPN", "NMOD", "PROPN"),
("NOUN", "NMOD", "PRON"),
("NOUN", "AMOD", "ADJ"),
("NOUN", "AMOD", "VERB"),
("PROPN", "AMOD", "ADJ"),
("PRON", "AMOD", "ADJ"),
("NOUN", "ACL", "VERB"),
("NOUN", "NUMMOD", "NUM"),
# Up to 85% of viruses that cause respiratory illness are identified by the technology.
("SYM", "NUMMOD", "NUM"),
("VERB", "ADVCL", "VERB"),
("ADJ", "ADVCL", "VERB"),
("VERB", "ADVCL", "ADJ"),
("VERB", "ADVCL", "NOUN"),
# The second memorable shift was in September, when the plant made the 75-millionth ton of steel.
("PROPN", "ADVCL", "VERB"),
# Although big city marathons offer great crowd support and a large camaraderie of runners, running in a big city marathon is not for everyone.
("PRON", "ADVCL", "VERB"),
# As the earth revolves around the sun, the place where light shines the brightest changes.
("NOUN", "ADVCL", "VERB"),
# We placed some wax into the old tin can.
("VERB", "ADVCL", "AUX"),
# I have no idea why this rather loud blazer from the GAP was in a bin at my local Dollar Tree.
("VERB", "ADVCL", "PROPN"),
}
self.mod_edges |= {
(pos1, dep, pos2)
for pos1 in self.npos
for pos2 in self.npos
for dep in ("NMOD", "AMOD")
}
def get_binary_fnc(self):
def r(edge):
return f'f_dep1(merge(merge(?2,"(r<root> :{edge} (d1<dep1>))"), r_dep1(?1)))' # noqa
coord = f'f_dep1(f_dep2(merge(merge(r_dep1(?1),"(coord<root> / COORD :0 (d1<dep1>) :0 (d2<dep2>))"), r_dep2(?2))))' # noqa
poss = f'f_relation(f_dep1(merge(merge(?2,"(has<relation> / HAS :2 (r<root>) :1 (d1<dep1>)))"), r_dep1(?1))))'
obl_tmod = f'f_relation(f_dep1(merge(merge(?2,"(at<relation> / AT :2 (r<root>) :1 (d1<dep1>)))"), r_dep1(?1))))'
csubj = f'f_dep1(merge(merge(?1,"(r<root> :1 (d1<dep1>))"), r_dep1(?2)))'
self.bin_fnc = {
("ADJ", "NSUBJ", "NOUN"): [r("1")],
("VERB", "NSUBJ_PASS", "NOUN"): [r("2")],
("VERB", "NSUBJ_PASS", "PRON"): [r("2")],
("VERB", "NSUBJ_PASS", "PROPN"): [r("2")],
# 38% of the world's generated electrical energy is gained from coal.
("VERB", "NSUBJ_PASS", "SYM"): [r("2")],
# CCOMP
("VERB", "CCOMP", "VERB"): [r("0")],
("VERB", "CCOMP", "ADJ"): [r("2")],
("VERB", "CCOMP", "ADJ"): [r("2")],
# make sure they have a copy of the invoice - sure ->CCOMP -> have
("ADJ", "CCOMP", "VERB"): [r("2")],
("VERB", "CCOMP", "NOUN"): [r("2")],
# It is no accident that the title of the exhibition is a homage to one of the classic figures of contemporaneity, Antoni Tapies, whose work breached all the boundaries imposed on artistic creation by the critics.
("NOUN", "CCOMP", "NOUN"): [r("2")],
# OBL
("VERB", "OBL", "NOUN"): [r("2")],
# Show 1 in the series is a documentary detailing the first stages of the celebrity students' conductor training as they enter into a week long 'Baton Camp'.
("VERB", "OBL", "PROPN"): [r("2")],
# Typically, the maximum heat generated from 24 fuel assemblies stored in a cask is less than that given off by a typical home heating system in an hour.
("ADJ", "OBL", "PRON"): [r("2")],
# Up to 85% of viruses that cause respiratory illness are identified by the technology.
("VERB", "OBL", "SYM"): [r("2")],
# The film makes the point that decision-making is an important aspect of such an affair of the heart.
("VERB", "OBL_TMOD", "NOUN"): [obl_tmod],
# ACL
# An Afghan handed over innocent people into torture. Afghan ADJ?
("PROPN", "ACL", "VERB"): [r("0")],
("ADJ", "ACL", "VERB"): [r("0")],
("NOUN", "ACL", "VERB"): [r("0")],
("PROPN", "ACL", "VERB"): [r("0")],
# The painting was one of the first used as a poster in an advertising campaign for soap powder.
("NUM", "ACL", "VERB"): [r("0")],
# The book asserts the notion that men and women are as different as beings from other planets.
("NOUN", "ACL", "ADJ"): [r("0")],
# It was a dispute discussing the question whether the language of the Greek people (Dimotiki) or a cultivated imitation of Ancient Greek (Katharevousa) should be the official language of the Greek nation.
("NOUN", "ACL", "NOUN"): [r("0")],
# FLAT - Andrew -flat> Wakefield
("PROPN", "FLAT", "PROPN"): [r("0")],
# Parataxis
("VERB", "PARATAXIS", "VERB"): [r("0")],
("NOUN", "PARATAXIS", "VERB"): [r("0")],
# 77793 civilians have arrived into the government controlled areas within the last two days.
("NUM", "PARATAXIS", "VERB"): [r("0")],
("PROPN", "PARATAXIS", "NOUN"): [r("0")],
# A hairdresser fine-tunes your hair color without causing excessive damage by using toners.
("ADJ", "PARATAXIS", "VERB"): [r("0")],
# The film makes the point that decision-making is an important aspect of such an affair of the heart.
("VERB", "PARATAXIS", "NOUN"): [r("0")],
# The 50g is hands-down, the absolute best calculator for engineers, surveyors, and hackers.
("NOUN", "PARATAXIS", "NOUN"): [r("0")],
# The image is from the poster 'Selling Counterfeit Products is Illegal'.
("NOUN", "PARATAXIS", "PROPN"): [r("0")],
# ACL_RELCL
("NOUN", "ACL_RELCL", "VERB"): [r("0")],
("PRON", "ACL_RELCL", "VERB"): [r("0")],
("NOUN", "ACL_RELCL", "ADJ"): [r("0")],
("PROPN", "ACL_RELCL", "VERB"): [r("0")],
("NOUN", "ACL_RELCL", "NOUN"): [r("0")],
("PROPN", "ACL_RELCL", "NOUN"): [r("0")],
# NSUBJ
("VERB", "NSUBJ", "NOUN"): [r("1")],
("VERB", "NSUBJ", "ADJ"): [r("1")],
("VERB", "NSUBJ", "PROPN"): [r("1")],
# One of the guards at the entrance of the supermarket said it had been the same scenario on Saturday with a long queue for sugar being the order of the day.
("VERB", "NSUBJ", "NUM"): [r("1")],
("VERB", "NSUBJ", "PRON"): [r("1")],
("ADJ", "NSUBJ", "PRON"): [r("1")],
# Ironically, the damage caused by the floods, and the subsequent insurance payout, were what prompted the restoration of the station building.
("PRON", "NSUBJ", "NOUN"): [r("1")],
("NOUN", "NSUBJ", "PRON"): [r("1")],
("NOUN", "NSUBJ", "PROPN"): [r("1")],
("ADJ", "NSUBJ", "NOUN"): [r("1")],
# Jennifer Holt, a former actress from Western movies, was "Aunt Judy," the only human in the cast. human ADJ?
("ADJ", "NSUBJ", "PROPN"): [r("1")],
("NOUN", "NSUBJ", "NOUN"): [r("1")],
("PROPN", "NSUBJ", "NOUN"): [r("1")],
("PROPN", "NSUBJ", "PRON"): [r("1")],
("ADV", "NSUBJ", "NOUN"): [r("1")],
# Here is Jose Mojica Marins, the popular Coffin Joe, a filmmaker who invented a cinema of total grossness.
("ADV", "NSUBJ", "PROPN"): [r("1")],
("ADV", "NSUBJ", "NUM"): [r("1")],
("VERB", "NSUBJ", "DET"): [r("1")],
# Patient survival one year after transplantation from a living-related donor is 95% and comparably high if the organ comes from a cadaveric donor. - For some reason % is the root
("SYM", "NSUBJ", "NOUN"): [r("1")],
# The painting was one of the first used as a poster in an advertising campaign for soap powder.
("NUM", "NSUBJ", "NOUN"): [r("1")],
("NUM", "NSUBJ", "PROPN"): [r("1")],
# CSUBJ
("ADJ", "CSUBJ", "VERB"): [csubj],
("NOUN", "CSUBJ", "VERB"): [csubj],
("VERB", "CSUBJ", "VERB"): [csubj],
("PROPN", "CSUBJ", "VERB"): [csubj],
# Although big city marathons offer great crowd support and a large camaraderie of runners, running in a big city marathon is not for everyone.
("PRON", "CSUBJ", "VERB"): [csubj],
("VERB", "CONJ", "VERB"): [coord],
("NOUN", "CASE", "ADP"): [r("0")],
("VERB", "XCOMP", "VERB"): [r("2")],
# make sure they have a copy of the invoice
("VERB", "XCOMP", "ADJ"): [r("2")],
# It becomes a lunch-time respite from the busy city life as a diner relaxes with his buffet delight while watching the commuters waiting alongside each other.
("VERB", "XCOMP", "NOUN"): [r("2")],
("ADJ", "XCOMP", "VERB"): [r("2")],
# It's called Symform - a stealthy Seattle outfit founded by a pair of ex-Microsoft employees, Praerit Garg and Bassam Tabbara.
("VERB", "XCOMP", "PROPN"): [r("2")],
# poss
("NOUN", "NMOD_POSS", "PRON"): [poss],
("NOUN", "NMOD_POSS", "PROPN"): [poss],
("NOUN", "NMOD_POSS", "NOUN"): [poss],
("PROPN", "NMOD_POSS", "PROPN"): [poss],
("PROPN", "NMOD_POSS", "PRON"): [poss],
# compound
("NOUN", "COMPOUND", "NOUN"): [r("0")],
("PROPN", "COMPOUND", "PROPN"): [r("0")],
("NOUN", "COMPOUND", "PROPN"): [r("0")],
("NUM", "COMPOUND", "NUM"): [r("0")],
# 77793 civilians have arrived into the government controlled areas within the last two days.
("VERB", "COMPOUND", "NOUN"): [r("0")],
("AUX", "COMPOUND", "NOUN"): [r("0")],
# obj
("VERB", "OBJ", "NOUN"): [r("2")],
("VERB", "OBJ", "PRON"): [r("2")],
("VERB", "OBJ", "PROPN"): [r("2")],
# Show 1 in the series is a documentary detailing the first stages of the celebrity students' conductor training as they enter into a week long 'Baton Camp'.
("VERB", "OBJ", "NUM"): [r("2")],
# The company builds many of the machines used in the manufacturing of the beds and the upfitting of the chassis.
("VERB", "OBJ", "ADJ"): [r("2")],
# Local fisherman heated some of their catch cooked over coals in a scuttle.
("VERB", "OBJ", "DET"): [r("2")],
# The company has made available its collection of luxurious full black parts.
("ADJ", "OBJ", "NOUN"): [r("2")],
# appos
("NOUN", "APPOS", "PROPN"): [r("0")],
("PROPN", "APPOS", "PROPN"): [r("0")],
# Dalindyebo R900-million and the tribe a further R80-billion in compensation for the humiliation caused by the monarch's criminal trial.
("PROPN", "APPOS", "NUM"): [r("0")],
# Jennifer Holt, a former actress from Western movies, was "Aunt Judy," the only human in the cast.
("PROPN", "APPOS", "ADJ"): [r("0")],
("NOUN", "APPOS", "NOUN"): [r("0")],
("PROPN", "APPOS", "NOUN"): [r("0")],
("VERB", "APPOS", "NOUN"): [r("0")],
# Non-UK students benefit from the bursary scheme.
("NOUN", "APPOS", "VERB"): [r("0")],
}
self.bin_fnc.update({edge: [r("0")] for edge in self.mod_edges})
# coordination
self.bin_fnc.update({("VERB", "CONJ", pos): [coord] for pos in self.npos})
self.bin_fnc.update({(pos, "CONJ", "VERB"): [coord] for pos in self.npos})
# I am a semi-professional myself and I also have a small hug of teddy bears, but the photos that Marc Hoberman made of these bears are really fantastic.
self.bin_fnc.update({("PRON", "CONJ", "VERB"): [coord]})
self.bin_fnc.update(
{(pos1, "CONJ", pos2): [coord] for pos1 in self.npos for pos2 in self.npos}
)
def get_term_fnc(self):
def n(label):
return self.get_default_terminal(label)
self.term_fnc = {
"not": [n("NEG")],
"none": [n("NEG")],
"no": [n("NEG")],
"non": [n("NEG")],
}
def handle_obl_case_mark(
self, parent_dep, parent_pos, current_pos, children_pos, i, j, clemma
):
if clemma == "by":
obl_case = (
f"{parent_pos} -> HANDLE_{parent_dep}_CASE_{children_pos}_{i}_{j}({children_pos}, {current_pos}, {parent_pos})",
{
"ud": f"{parent_pos}_2(_{parent_dep}_1({current_pos}_2(_CASE_1(?1), ?2)),?3)",
"fl": f'f_dep1(merge(merge(?3,"(r<root> :1 (d1<dep1> :0 (r<root>)))"), r_dep1(?2)))',
},
"nonterminal",
)
else:
obl_case = (
f"{parent_pos} -> HANDLE_{parent_dep}_CASE_{children_pos}_{i}_{j}({children_pos}, {current_pos}, {parent_pos})",
{
"ud": f"{parent_pos}_2(_{parent_dep}_1({current_pos}_2(_CASE_1(?1), ?2)),?3)",
"fl": f'f_dep2(f_dep1(merge(merge(merge(?3,"(d1<dep1> :1 (r<root>) :2 (d2<dep2>))"), r_dep1(?1)), r_dep2(?2))))',
},
"nonterminal",
)
# The opposition leader has gone into hiding.
obl_mark = (
f"{parent_pos} -> HANDLE_{parent_dep}_MARK_{children_pos}_{i}_{j}({children_pos}, {current_pos}, {parent_pos})",
{
"ud": f"{parent_pos}_2(_{parent_dep}_1({current_pos}_2(_MARK_1(?1), ?2)),?3)",
"fl": f'f_dep2(f_dep1(merge(merge(merge(?3,"(d1<dep1> :1 (r<root>) :2 (d2<dep2>))"), r_dep1(?1)), r_dep2(?2))))',
},
"nonterminal",
)
return [obl_case, obl_mark]
def handle_advcl_mark(
self, parent_dep, parent_pos, current_pos, children_pos, i, j, clemma
):
advcl_mark = (
f"{parent_pos} -> HANDLE_{parent_dep}_MARK_{children_pos}_{i}_{j}({children_pos}, {current_pos}, {parent_pos})",
{
"ud": f"{parent_pos}_2(_{parent_dep}_1({current_pos}_2(_MARK_1(?1), ?2)),?3)",
"fl": f'f_dep2(f_dep1(merge(merge(merge(?3,"(d1<dep1> :1 (r<root>) :2 (d2<dep2>))"), r_dep1(?1)), r_dep2(?2))))',
},
"nonterminal",
)
return [advcl_mark]
def handle_acl_relcl(self, dep, parent_pos, current_pos, children_pos, i, j):
acl_relcl = (
f"{parent_pos} -> HANDLE_ACL_RELCL_NSUBJ_{children_pos}_{i}_{j}({children_pos}, {current_pos}, {parent_pos})",
{
"ud": f"{parent_pos}_2(_ACL_RELCL_1({current_pos}_2(_NSUBJ_1(?1), ?2)),?3)",
"fl": f'f_dep1(merge(merge(?3,"(r<root> :0 (d1<dep1> :1 (r<root>)))"), r_dep1(?2)))',
},
"nonterminal",
)
acl_relcl2 = (
f"{parent_pos} -> HANDLE_ACL_RELCL_NSUBJ_PASS_{children_pos}_{i}_{j}({children_pos}, {current_pos}, {parent_pos})",
{
"ud": f"{parent_pos}_2(_ACL_RELCL_1({current_pos}_2(_NSUBJ_PASS_1(?1), ?2)),?3)",
"fl": f'f_dep1(merge(merge(?3,"(r<root> :0 (d1<dep1> :1 (r<root>)))"), r_dep1(?2)))',
},
"nonterminal",
)
return [acl_relcl, acl_relcl2]
def handle_subgraphs(self, lemma, pos, clemma, cpos, dep, parent, i, j):
parent_dep = parent[2]
parent_pos = parent[1]
if parent_dep in ("NMOD", "OBL", "OBL_NPMOD"):
return self.handle_obl_case_mark(
parent_dep, parent_pos, pos, cpos, i, j, clemma
)
if parent_dep in ("ADVCL"):
return self.handle_advcl_mark(
parent_dep, parent_pos, pos, cpos, i, j, clemma
)
if parent_dep == "ACL_RELCL":
return self.handle_acl_relcl(dep, parent_pos, pos, cpos, i, j)
return
class CFLLexicon(BaseLexicon):
def __init__(self):
super(CFLLexicon, self).__init__()
def get_mod_edges(self):
self.mod_edges = {
("ADJ", "ADVMOD", "ADV"),
# nicht hoeher, 7774_18_1
("ADJ", "ADVMOD", "PART"),
# tatsaechlich errichteten, 7774_18_1
("ADJ", "ADVMOD", "ADJ"),
# zulaessig -> bis, sample 6
("ADJ", "CASE", "ADP"),
# bis -> Ausladung, sample 6
("ADP", "NMOD", "NOUN"),
# Raum darueber, 318 of sample_10
("NOUN", "ADVMOD", "ADV"),
("NOUN", "ADVMOD", "NUM"),
# kein Dachgaube, sample 4
("NOUN", "ADVMOD", "PRON"),
# Gebaeudefront_NOUN -AMOD-> bzw_VERB, sample 11
("NOUN", "AMOD", "VERB"),
# Gebaeudefront_NOUN -ACL-> liegen_VERB, sample 5
("NOUN", "ACL", "VERB"),
# Aussenminister ... der strunzdumm ist
("NOUN", "ACL", "ADJ"),
("NOUN", "NUMMOD", "NUM"),
("NUM", "ADVMOD", "ADV"),
# maximal 6,0 m
("NUM", "ADVMOD", "ADJ"),
("NUM", "ADVMOD", "NUM"),
("VERB", "ADVMOD", "ADJ"),
("VERB", "ADVMOD", "ADV"),
# nicht staffeln, sample 10
("VERB", "ADVMOD", "PART"),
# nicht mehr als
("CCONJ", "ADVMOD", "ADV"),
# nicht mehr als
("ADV", "ADVMOD", "PART"),
# sample 112 of sample_10
("VERB", "ADVCL", "VERB"),
# nicht gewaehlt... , weil er gegen die Homo-Ehe... (Germeval '18)
("VERB", "ADVCL", "ADJ"),
# betragen duerfen, sample 13
("VERB", "AUX", "AUX"),
# liegen -> Baulinie, sample 6
# ("VERB", "OBL", "NOUN"),
}
self.mod_edges |= {
(pos1, dep, pos2)
for pos1 in self.npos
for pos2 in self.npos
for dep in ("NMOD", "AMOD")
}
def get_binary_fnc(self):
def r(edge):
return f'f_dep1(merge(merge(?2,"(r<root> :{edge} (d1<dep1>))"), r_dep1(?1)))' # noqa
coord = f'f_dep1(f_dep2(merge(merge(r_dep1(?1),"(coord<root> / COORD :0 (d1<dep1>) :0 (d2<dep2>))"), r_dep2(?2))))' # noqa
self.bin_fnc = {
# Errichtung ist untersagt
("ADJ", "NSUBJ", "NOUN"): [r("1")],
# Verloren ist die Zeit
("NOUN", "NSUBJ", "NOUN"): [r("1")],
("PROPN", "NSUBJ", "NOUN"): [r("1")],
("NOUN", "NSUBJ", "PROPN"): [r("1")],
("VERB", "NSUBJ_PASS", "NOUN"): [r("2")],
# jeder der ... paktiert, hat ... verschissen
("VERB", "CSUBJ", "VERB"): [r("1")],
# ...wird bestimmt, dass...
("VERB", "CSUBJ_PASS", "VERB"): [r("2")],
# ...wird bestimmt, dass...
("VERB", "CCOMP", "VERB"): [r("2")],
# ...wird bestimmt: Die Errichtung...zulaesssig (8159_21_0)
# ...wird bestimmt: ...betragen
# parsed as parataxis
("VERB", "PARATAXIS", "ADJ"): [r("2")],
("VERB", "PARATAXIS", "VERB"): [r("2")],
# ...so auszubilden, dass...
("VERB", "CCOMP", "ADJ"): [r("0")], # TODO
# sind Vorkehrungen zu treffen, dass...moeglich
("NOUN", "CCOMP", "ADJ"): [r("0")], # TODO
# sind Vorkehrungen zu treffen, dass...bleiben
("NOUN", "CCOMP", "VERB"): [r("0")], # TODO
# vorhanden bleiben (correct parse? why not obj or obl?)
("VERB", "XCOMP", "ADJ"): [r("2")], # TODO
# hat ... zu tun (Germeval '18)
("VERB", "XCOMP", "VERB"): [r("2")], # TODO
("VERB", "OBJ", "NOUN"): [r("2")],
# habt ihr Angst
("AUX", "OBJ", "NOUN"): [r("2")],
("NOUN", "OBJ", "NOUN"): [r("2")],
("VERB", "OBJ", "PROPN"): [r("2")],
# hat nichts... (Germeval '18)
("VERB", "OBJ", "PRON"): [r("2")],
# Fuer alle Flaechen ... zu treffen # TODO
("VERB", "OBL", "NOUN"): [r("2")],
# zu begruenen, e.g. 7181_6_0
("VERB", "MARK", "PART"): [r("0")],
# sofern.., e.g. 7408_10_1
("VERB", "MARK", "SCONJ"): [r("0")],
("VERB", "NSUBJ", "NOUN"): [r("1")],
# ...Pflanzung möglich ist...
("VERB", "NSUBJ", "ADJ"): [r("1")],
("VERB", "NSUBJ", "PRON"): [r("1")],
("VERB", "NSUBJ", "PROPN"): [r("1")],
("VERB", "CONJ", "VERB"): [coord],
("VERB", "CONJ", "AUX"): [coord],
("NOUN", "CASE", "ADP"): [r("0")],
# 7181_3_1
("NOUN", "APPOS", "PROPN"): [r("0")],
("PROPN", "APPOS", "PROPN"): [r("0")],
("PROPN", "FLAT", "PROPN"): [r("0")],
("NOUN", "COMPOUND", "NOUN"): [r("0")],
# Rede ... jetzt
("NOUN", "APPOS", "ADV"): [r("0")],
}
self.bin_fnc.update({edge: [r("0")] for edge in self.mod_edges})
# coordination
self.bin_fnc.update({("VERB", "CONJ", pos): [coord] for pos in self.npos})
self.bin_fnc.update({(pos, "CONJ", "VERB"): [coord] for pos in self.npos})
self.bin_fnc.update(
{(pos1, "CONJ", pos2): [coord] for pos1 in self.npos for pos2 in self.npos}
)
def get_term_fnc(self):
def n(label):
return self.get_default_terminal(label)
self.term_fnc = {
"nicht": [n("NEG")],
"kein": [n("NEG")],
"duerfen": [n("PER")],
"muessen": [n("OBL")],
"zulaessig": [n("PER")],
"sofern": [n("EXC")],
"untersagen": [n("FOR")],
"unzulaessig": [n("FOR")],
("zu", "PART"): [n("OBL")],
}
def handle_obl_case(
self, parent_dep, parent_pos, current_pos, children_pos, i, clemma
):
obl_case = (
f"{parent_pos} -> HANDLE_{parent_dep}_CASE_{children_pos}_{i}({children_pos}, {current_pos}, {parent_pos})",
{
"ud": f"{parent_pos}_2(_{parent_dep}_1({current_pos}_2(_CASE_1(?1), ?2)),?3)",
"fl": f'f_dep2(f_dep1(merge(merge(merge(?3,"(d1<dep1> :1 (r<root>) :2 (d2<dep2>))"), r_dep1(?1)), r_dep2(?2))))',
},
"nonterminal",
)
return [obl_case]
def handle_subgraphs(self, lemma, pos, clemma, cpos, dep, parent, i, j):
parent_dep = parent[2]
parent_pos = parent[1]
if parent_dep in ("NMOD", "OBL"):
return self.handle_obl_case(parent_dep, parent_pos, pos, cpos, i, clemma)
return