-
Notifications
You must be signed in to change notification settings - Fork 8
/
Copy pathfull_text_reader.py
493 lines (448 loc) · 24.2 KB
/
full_text_reader.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
import codecs
import logging
import os
import xml.etree.ElementTree as ElementTree
from typing import Any, Dict, List, Optional, Tuple, DefaultDict, Set
import tqdm
from overrides import overrides
from allennlp.common import Params
from allennlp.common.checks import ConfigurationError
from allennlp.data.dataset import Dataset
from allennlp.data.dataset_readers.dataset_reader import DatasetReader
from allennlp.data.dataset_readers.framenet.ontology_reader import FrameOntology
from allennlp.data.fields import Field, IndexField, ListField, SequenceLabelField, TextField, LabelField
from allennlp.data.instance import Instance
from allennlp.data.token_indexers import SingleIdTokenIndexer, TokenIndexer
from allennlp.data.tokenizers import Token
logger = logging.getLogger(__name__) # pylint: disable=invalid-name
@DatasetReader.register("fn_full")
class FrameNetFullTextReader(DatasetReader):
"""
This DatasetReader is designed to read the FrameNet 1.x full text data in xml format.
fndata-1.5/
├── development
│ └── fulltext
│ ├── ANC__110CYL072.xml
│ ├── KBEval__MIT.xml
│ ├── LUCorpus-v0.3__20000415_apw_eng-NEW.xml
│ ├── LUCorpus-v0.3__ENRON-pearson-email-25jul02.xml
│ ├── Miscellaneous__Hijack.xml
│ ├── NTI__NorthKorea_NuclearOverview.xml
│ ├── NTI__WMDNews_062606.xml
│ └── PropBank__TicketSplitting.xml
├── frame -> /homes/gws/swabha/acl2017/data/fndata-1.5/frame
├── frameIndex.xml -> /homes/gws/swabha/acl2017/data/fndata-1.5/frameIndex.xml
├── luIndex.xml -> /homes/gws/swabha/acl2017/data/fndata-1.5/luIndex.xml
├── test
│ └── fulltext
│ ├── ANC__110CYL067.xml
│ ├── ANC__110CYL069.xml
│ ├── ANC__112C-L013.xml
│ ├── ANC__IntroHongKong.xml
│ ├── ANC__StephanopoulosCrimes.xml
│ ├── ANC__WhereToHongKong.xml
│ ├── KBEval__atm.xml
│ ├── KBEval__Brandeis.xml
│ ├── KBEval__cycorp.xml
│ ├── KBEval__parc.xml
│ ├── KBEval__Stanford.xml
│ ├── KBEval__utd-icsi.xml
│ ├── LUCorpus-v0.3__20000410_nyt-NEW.xml
│ ├── LUCorpus-v0.3__AFGP-2002-602187-Trans.xml
│ ├── LUCorpus-v0.3__enron-thread-159550.xml
│ ├── LUCorpus-v0.3__IZ-060316-01-Trans-1.xml
│ ├── LUCorpus-v0.3__SNO-525.xml
│ ├── LUCorpus-v0.3__sw2025-ms98-a-trans.ascii-1-NEW.xml
│ ├── Miscellaneous__Hound-Ch14.xml
│ ├── Miscellaneous__SadatAssassination.xml
│ ├── NTI__NorthKorea_Introduction.xml
│ ├── NTI__Syria_NuclearOverview.xml
│ └── PropBank__AetnaLifeAndCasualty.xml
└── train
├── exemplar-train -> /homes/gws/swabha/acl2017/data/fndata-1.5/lu
└── fulltext
├── ANC__110CYL068.xml
├── ANC__110CYL070.xml
├── ANC__110CYL200.xml
├── ANC__EntrepreneurAsMadonna.xml
├── ANC__HistoryOfGreece.xml
├── ANC__HistoryOfJerusalem.xml
├── ANC__HistoryOfLasVegas.xml
├── ANC__IntroJamaica.xml
├── ANC__IntroOfDublin.xml
├── C-4__C-4Text.xml
├── KBEval__lcch.xml
├── KBEval__LCC-M.xml
├── LUCorpus-v0.3__20000416_xin_eng-NEW.xml
├── LUCorpus-v0.3__20000419_apw_eng-NEW.xml
├── LUCorpus-v0.3__20000420_xin_eng-NEW.xml
├── LUCorpus-v0.3__20000424_nyt-NEW.xml
├── LUCorpus-v0.3__602CZL285-1.xml
├── LUCorpus-v0.3__AFGP-2002-600002-Trans.xml
├── LUCorpus-v0.3__AFGP-2002-600045-Trans.xml
├── LUCorpus-v0.3__artb_004_A1_E1_NEW.xml
├── LUCorpus-v0.3__artb_004_A1_E2_NEW.xml
├── LUCorpus-v0.3__CNN_AARONBROWN_ENG_20051101_215800.partial-NEW.xml
├── LUCorpus-v0.3__CNN_ENG_20030614_173123.4-NEW-1.xml
├── LUCorpus-v0.3__wsj_1640.mrg-NEW.xml
├── LUCorpus-v0.3__wsj_2465.xml
├── NTI__BWTutorial_chapter1.xml
├── NTI__ChinaOverview.xml
├── NTI__Iran_Biological.xml
├── NTI__Iran_Chemical.xml
├── NTI__Iran_Introduction.xml
├── NTI__Iran_Missile.xml
├── NTI__Iran_Nuclear.xml
├── NTI__Kazakhstan.xml
├── NTI__LibyaCountry1.xml
├── NTI__NorthKorea_ChemicalOverview.xml
├── NTI__NorthKorea_NuclearCapabilities.xml
├── NTI__Russia_Introduction.xml
├── NTI__SouthAfrica_Introduction.xml
├── NTI__Taiwan_Introduction.xml
├── NTI__WMDNews_042106.xml
├── NTI__workAdvances.xml
├── PropBank__BellRinging.xml
├── PropBank__ElectionVictory.xml
├── PropBank__LomaPrieta.xml
├── PropBank__PolemicProgressiveEducation.xml
├── QA__IranRelatedQuestions.xml
└── SemAnno__Text1.xml
"""
def __init__(self,
max_span_width: int,
data_path: str,
token_indexers: Dict[str, TokenIndexer] = None,
frame_indexers: Dict[str, TokenIndexer] = None,
syn_indexers: Dict[str, TokenIndexer] = None) -> None:
self._max_span_width = max_span_width
self._token_indexers = token_indexers or {"tokens": SingleIdTokenIndexer()}
self._frame_indexers = frame_indexers or {"frames": SingleIdTokenIndexer()}
self._syn_indexers = syn_indexers or {"syn_labels": SingleIdTokenIndexer()}
self._ontology = FrameOntology(data_path)
self._namespace = {"fn": "http://framenet.icsi.berkeley.edu"}
self._tokenization_layers = ["BNC", "PENN"]
self._num_sents = 0
self._valid_sents = 0
self._total_sentence_length = 0
self._discontinuous_targets = 0
self._adjacent_labeled_args = 0
self._total_labeled_args = 0
def _reset(self):
self._num_sents = 0
self._valid_sents = 0
self._total_sentence_length = 0
self._discontinuous_targets = 0
self._adjacent_labeled_args = 0
self._total_labeled_args = 0
def fill_missing_spans(self, fe_list: List[Tuple[int, int, str]], sentence_length: int):
"""
Also counts adjacent spans wiht same labels.
"""
if not fe_list:
return [(0, sentence_length-1, "O")]
prev_label = None
prev_end = -1
args = sorted([x for x in fe_list])
for arg in args:
start, end, label = arg
if prev_end != start - 1:
fe_list.append((prev_end + 1, start-1, "O"))
else:
if prev_label == label:
self._adjacent_labeled_args += 1
prev_end = end
prev_label = label
if prev_end != sentence_length-1:
fe_list.append((prev_end+1, sentence_length-1, "O"))
return sorted(fe_list)
def _convert_frame_elements_to_matrix(self, fe_list: List[Tuple[int, int, str]], sentence_length: int) -> List[List[str]]:
matrix = [["*" for _ in range(self._max_span_width)]
for _ in range(sentence_length)]
self._total_labeled_args += len(fe_list)
fe_map_refilled = self.fill_missing_spans(fe_list, sentence_length)
for span in fe_map_refilled:
start, end, label = span
diff = end - start
while diff >= self._max_span_width:
matrix[end][self._max_span_width - 1] = label
end = end - self._max_span_width
diff = end - start
matrix[end][diff] = label
return matrix
def _convert_labeled_spans_to_sequence(self, spans: List[Tuple[int, int, str]], sequence_len: int) -> List[str]:
sequence = ["-"] * sequence_len
for span in spans:
start, end, label = span
for position in range(start, end+1):
assert sequence[position] == "-"
sequence[position] = label
return sequence
def _process_sentence(self,
sentence_tokens: List[str],
targets: List[List[Tuple[int, int]]],
lexical_units: List[str],
frames: List[str],
frame_elements: List[List[Tuple[int, int, str]]],
phrase_types: List[List[Tuple[int, int, str]]],
grammar_functions: List[List[Tuple[int, int, str]]]) -> List[Instance]:
instances = []
for target, lex_unit, frame, fe, pt, gf in zip(targets, lexical_units, frames, frame_elements, phrase_types, grammar_functions):
target_label = [0 for _ in sentence_tokens]
if len(target) > 1:
self._discontinuous_targets += 1
for target_span in target:
for idx in range(target_span[0], target_span[1] + 1):
assert target_label[idx] == 0
target_label[idx] = 1
fe_matrix = self._convert_frame_elements_to_matrix(fe, len(sentence_tokens))
# For simplicity, we assume the target index to be the first target position.
target_index = target[0][0]
pt_seq = self._convert_labeled_spans_to_sequence(pt, len(sentence_tokens))
gf_seq = self._convert_labeled_spans_to_sequence(gf, len(sentence_tokens))
instances.append(self.text_to_instance(
sentence_tokens, target_label, target_index, lex_unit, frame, fe_matrix, pt_seq, gf_seq))
self._valid_sents += 1
self._total_sentence_length += len(sentence_tokens)
return instances
@overrides
def read(self, file_path: str):
logger.info(
"Reading FrameNet full text instances from dataset files at: %s", file_path)
instances = []
# prev_len = len(instances)
for root, _, directory in tqdm.tqdm(list(os.walk(file_path))):
for data_file in sorted(directory):
if not data_file.endswith(".xml"):
continue
instances.extend(self.read_single_fulltext_file(
os.path.join(root, data_file)))
# logger.info("%s: # instances = %d", data_file, len(instances) - prev_len)
# prev_len = len(instances)
logger.info("# instances = %d", len(instances))
logger.info("# sentences = %d", self._num_sents)
logger.info("# valid sentences = %d", self._valid_sents)
logger.info("# avg tokens in sentence = %f",
self._total_sentence_length/self._valid_sents)
logger.info("# discontinuous targets = %d",
self._discontinuous_targets)
logger.info("%% adjacent spans with same label = %f (%d/%d)", self._adjacent_labeled_args /
self._total_labeled_args, self._adjacent_labeled_args, self._total_labeled_args)
self._reset()
return Dataset(instances)
def read_single_fulltext_file(self, data_file: str):
instances = []
with codecs.open(data_file, "rb", "utf-8") as xml_file:
tree = ElementTree.parse(xml_file)
root = tree.getroot()
full_text_filename = data_file.split("/")[-1]
is_test_file = "test" in data_file
for sentence in root.findall("fn:sentence", self._namespace):
tokens: List[str] = []
pos_tags: List[str] = []
starts: Dict[int, int] = {}
ends: Dict[int, int] = {}
targets: List[List[Tuple[int, int]]] = []
lexical_units: List[str] = []
frames: List[str] = []
frame_elements: List[List[Tuple[int, int, str]]] = []
phrase_types: List[List[Tuple[int, int, str]]] = []
grammar_functs: List[List[Tuple[int, int, str]]] = []
sentence_text = sentence.find("fn:text", self._namespace).text
self._num_sents += 1
annotations = sentence.findall("fn:annotationSet", self._namespace)
# An annotation could either contain the tokenization for the sentence or a frame-semantic parse.
for annotation in annotations:
if "luName" in annotation.attrib and "frameName" in annotation.attrib:
# Ignore the unannotated instances in ONLY dev/train.
if annotation.attrib["status"] == "UNANN" and not is_test_file:
continue
# Get the LU, Frame and FEs for this sentence.
lex_unit = annotation.attrib["luName"]
frame = annotation.attrib["frameName"]
if frame == "Test35": # Bogus frame.
continue
target_tokens = []
frame_element_list = []
phrase_type_list = []
grammar_funct_list = []
# Targets and frame-elements.
for layer in annotation.findall("fn:layer", self._namespace):
layer_type = layer.attrib["name"]
if layer_type == "Target":
# Recover the target span.
target_labels = layer.findall("fn:label", self._namespace)
# Some annotations have missing targets - ignore those.
if not target_labels:
logger.info("Skipping: Missing target label at %s in %s",
annotation.attrib["ID"], full_text_filename)
break
# There can be discontinous targets.
for label in target_labels:
try:
start_token = starts[int(label.attrib["start"])]
end_token = ends[int(label.attrib["end"])]
except:
logger.info("Skipping: Start/End labels missing for target annotation %s in %s",
annotation.attrib["ID"], full_text_filename)
continue
target_tokens.append((start_token, end_token))
elif layer.attrib["name"] == "FE" and layer.attrib["rank"] == "1":
# Recover the frame elements.
for label in layer.findall("fn:label", self._namespace):
if "itype" in label.attrib:
continue
try:
start_token = starts[int(label.attrib["start"])]
end_token = ends[int(label.attrib["end"])]
frame_element_list.append((start_token, end_token, label.attrib["name"]))
except:
logger.info("Skipping: Frame-elements annotated for missing tokenization at annotation %s in %s",
annotation.attrib["ID"], full_text_filename)
continue
elif layer.attrib["name"] == "GF":
for label in layer.findall("fn:label", self._namespace):
try:
start_token = starts[int(label.attrib["start"])]
end_token = ends[int(label.attrib["end"])]
grammar_funct_list.append((start_token, end_token, label.attrib["name"]))
except:
continue
elif layer.attrib["name"] == "PT":
for label in layer.findall("fn:label", self._namespace):
try:
start_token = starts[int(label.attrib["start"])]
end_token = ends[int(label.attrib["end"])]
phrase_type_list.append((start_token, end_token, label.attrib["name"]))
except:
continue
if not target_tokens:
logger.info("Skipping: Missing target in annotation %s in %s", annotation.attrib["ID"], full_text_filename)
continue
if target_tokens in targets:
logger.info("Skipping: Repeated annotation %s for frame %s in %s", annotation.attrib["ID"], frame, full_text_filename)
continue
lexical_units.append(lex_unit)
frames.append(frame)
targets.append(target_tokens)
frame_elements.append(frame_element_list)
grammar_functs.append(grammar_funct_list)
phrase_types.append(phrase_type_list)
else:
for layer in annotation.findall("fn:layer", self._namespace):
if layer.attrib["name"] not in self._tokenization_layers:
continue
tokenization = {}
for label in layer.findall("fn:label", self._namespace):
start = int(label.attrib["start"])
end = int(label.attrib["end"])
tokenization[(start, end)] = label.attrib["name"]
previous_end = -2
for start_end in sorted(tokenization):
start, end = start_end
if start != previous_end + 2:
logger.info(
"Fixing: Missing tokenization at annotation %s in %s.", annotation.attrib["ID"], full_text_filename)
# Creating a new token.
dummy_start = previous_end + 2
dummy_end = start - 2
tokens.append(
sentence_text[dummy_start: dummy_end + 1])
pos_tags.append("missing")
starts[dummy_start] = len(tokens) - 1
ends[dummy_end] = len(tokens) - 1
tokens.append(sentence_text[start: end + 1])
pos_tags.append(tokenization[start_end])
starts[start] = len(tokens) - 1
ends[end] = len(tokens) - 1
previous_end = end
break
if not targets:
# Sentence with missing target annotations, will be skipped.
continue
assert len(pos_tags) == len(tokens)
assert len(targets) == len(lexical_units) == len(frames) == len(frame_elements)
instances.extend(self._process_sentence(sentence_tokens=tokens,
targets=targets,
lexical_units=lexical_units,
frames=frames,
frame_elements=frame_elements,
phrase_types=phrase_types,
grammar_functions=grammar_functs))
xml_file.close()
if not instances:
raise ConfigurationError("No instances were read from the given filepath {}. "
"Is the path correct?".format(data_file))
return instances
@overrides
def text_to_instance(self,
tokens: List[str],
target_label: List[int],
target_index: int,
lex_label: str,
frame_label: str,
frame_elements: List[List[str]] = None,
phrase_types: List[str] = None,
grammar_functs: List[str] = None) -> Instance:
# pylint: disable=arguments-differ
text_field = TextField([Token(t) for t in tokens], token_indexers=self._token_indexers)
verb_field = SequenceLabelField(target_label, text_field)
target_field = IndexField(target_index, text_field)
frame_field = TextField([Token(frame_label)], token_indexers=self._frame_indexers)
# lex_field = TextField([Token(lex_label)], token_indexers=self._token_indexers)
phrase_type_field = TextField([Token(p) for p in phrase_types], token_indexers=self._syn_indexers)
grammar_funct_field = TextField([Token(gf) for gf in grammar_functs], token_indexers=self._syn_indexers)
# Span-based output fields.
span_starts: List[Field] = []
span_ends: List[Field] = []
# span_mask: List[int] = [1 for _ in range(len(tokens) * self._max_span_width)]
span_labels: Optional[List[str]] = [] if frame_elements is not None else None
for j in range(len(tokens)):
for diff in range(self._max_span_width):
width = diff
if j - diff < 0:
# This is an invalid span.
# span_mask[j * self._max_span_width + diff] = 0
width = j
span_starts.append(IndexField(j - width, text_field))
span_ends.append(IndexField(j, text_field))
if frame_elements is not None:
current_label = frame_elements[j][diff]
span_labels.append(current_label)
start_fields = ListField(span_starts)
end_fields = ListField(span_ends)
# span_mask_fields = SequenceLabelField(span_mask, start_fields)
# Valid labels for frame.
valid_frame_elements = ListField([LabelField(x) for x in self._ontology.frame_fe_map[frame_label] + ["O"]])
fields: Dict[str, Field] = {'tokens': text_field,
'targets': verb_field,
'target_index': target_field,
# 'lu': lex_field,
'frame': frame_field,
'span_starts': start_fields,
'span_ends': end_fields,
# 'span_mask': span_mask_fields,
'valid_frame_elements': valid_frame_elements}
if frame_elements:
fields['tags'] = SequenceLabelField(span_labels, start_fields)
if grammar_functs:
fields['gf'] = grammar_funct_field
if phrase_types:
fields['pt'] = phrase_type_field
return Instance(fields)
@classmethod
def from_params(cls, params: Params) -> 'FrameNetFullTextReader':
token_indexers = TokenIndexer.dict_from_params(
params.pop('token_indexers', {}))
frame_indexers = TokenIndexer.dict_from_params(
params.pop('frame_indexers', {"frames": {"type": "single_id", "namespace": "frames"}}))
syn_indexers = TokenIndexer.dict_from_params(
params.pop('syn_indexers', {"syn_labels": {"type": "single_id", "namespace": "syn_labels"}}))
max_span_width = params.pop('max_span_width')
data_path = params.pop('data_path', None)
params.assert_empty(cls.__name__)
return cls(token_indexers=token_indexers,
frame_indexers=frame_indexers,
syn_indexers=syn_indexers,
max_span_width=max_span_width,
data_path=data_path)