Albert Sawczyn commited on
Commit
b194823
1 Parent(s): 5b08b7f

add dataset

Browse files
.gitattributes CHANGED
@@ -14,3 +14,4 @@
14
  *.pb filter=lfs diff=lfs merge=lfs -text
15
  *.pt filter=lfs diff=lfs merge=lfs -text
16
  *.pth filter=lfs diff=lfs merge=lfs -text
 
 
14
  *.pb filter=lfs diff=lfs merge=lfs -text
15
  *.pt filter=lfs diff=lfs merge=lfs -text
16
  *.pth filter=lfs diff=lfs merge=lfs -text
17
+ *.iob filter=lfs diff=lfs merge=lfs -text
README.md ADDED
@@ -0,0 +1,402 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ annotations_creators:
3
+ - expert-generated
4
+ language_creators:
5
+ - found
6
+ languages:
7
+ - pl
8
+ licenses:
9
+ - cc-by-3.0
10
+ multilinguality:
11
+ - monolingual
12
+ pretty_name: ''
13
+ size_categories:
14
+ - 18K
15
+ - 10K<n<100K
16
+ source_datasets:
17
+ - original
18
+ task_categories:
19
+ - structure-prediction
20
+ task_ids:
21
+ - named-entity-recognition
22
+ ---
23
+
24
+ # KPWR-NER
25
+
26
+ ## Description
27
+
28
+ KPWR-NER is a part the Polish Corpus of Wrocław University of Technology (*Korpus Języka Polskiego Politechniki Wrocławskiej*). Its objective is named entity recognition for fine-grained categories of entities. It is the ‘n82’ version of the KPWr, which means that number of classes is restricted to 82 (originally 120). During corpus creation, texts were annotated by humans from various sources, covering many domains and genres.
29
+
30
+ ## Tasks (input, output and metrics)
31
+
32
+ Named entity recognition (NER) - tagging entities in text with their corresponding type.
33
+
34
+ Input ('*tokens'* column): sequence of tokens
35
+
36
+ Output ('*ner'* column): sequence of predicted tokens’ classes in BIO notation (82 possible classes, described in detail in the annotation guidelines)
37
+
38
+ *example:*
39
+
40
+ [*‘Roboty’, ‘mają’, ‘kilkanaście’, ‘lat’, ‘i’, ‘pochodzą’, ‘z’, ‘USA’, ‘,’, ‘Wysokie’, ‘napięcie’, ‘jest’, ‘dużo’, ‘młodsze’, ‘,’, ‘powstało’, ‘w’, ‘Niemczech’, ‘.’*] → [*‘B-nam_pro_title’, ‘O’, ‘O’, ‘O’, ‘O’, ‘O’, ‘O’, ‘B-nam_loc_gpe_country’, ‘O’, ‘B-nam_pro_title’, ‘I-nam_pro_title’, ‘O’, ‘O’, ‘O’, ‘O’, ‘O’, ‘O’, ‘B-nam_loc_gpe_country’, ‘O’*]
41
+
42
+ Measurements:
43
+
44
+ ## Data splits
45
+
46
+
47
+ | Subset | Cardinality (sentences) |
48
+ |--------|------------------------:|
49
+ | train | 13959 |
50
+ | test | 4323 |
51
+
52
+ ## Class distribution in train
53
+
54
+ | Class | Fraction of tokens |
55
+ |:----------------------------|---------------------:|
56
+ | O | 0.898080 |
57
+ | B-nam_liv_person | 0.012769 |
58
+ | I-nam_liv_person | 0.008246 |
59
+ | I-nam_org_institution | 0.006448 |
60
+ | B-nam_loc_gpe_city | 0.005886 |
61
+ | B-nam_loc_gpe_country | 0.004351 |
62
+ | I-nam_org_organization | 0.003728 |
63
+ | B-nam_org_institution | 0.003434 |
64
+ | I-nam_pro_title_document | 0.003206 |
65
+ | B-nam_org_organization | 0.002592 |
66
+ | B-nam_org_group_team | 0.002035 |
67
+ | B-nam_adj_country | 0.001987 |
68
+ | I-nam_org_group_team | 0.001851 |
69
+ | I-nam_pro_title | 0.001715 |
70
+ | I-nam_eve_human | 0.001557 |
71
+ | B-nam_org_company | 0.001421 |
72
+ | I-nam_org_company | 0.001316 |
73
+ | B-nam_pro_media_periodic | 0.001312 |
74
+ | I-nam_fac_goe | 0.001285 |
75
+ | I-nam_pro_media_periodic | 0.001254 |
76
+ | B-nam_fac_road | 0.001162 |
77
+ | B-nam_liv_god | 0.001127 |
78
+ | I-nam_eve_human_sport | 0.001057 |
79
+ | B-nam_org_nation | 0.001013 |
80
+ | B-nam_oth_tech | 0.001004 |
81
+ | B-nam_pro_media_web | 0.000996 |
82
+ | B-nam_fac_goe | 0.000930 |
83
+ | B-nam_eve_human | 0.000917 |
84
+ | B-nam_pro_title | 0.000908 |
85
+ | B-nam_pro_brand | 0.000899 |
86
+ | I-nam_pro_model_car | 0.000877 |
87
+ | I-nam_pro_brand | 0.000868 |
88
+ | I-nam_loc_gpe_city | 0.000847 |
89
+ | B-nam_org_political_party | 0.000737 |
90
+ | I-nam_loc_gpe_country | 0.000715 |
91
+ | B-nam_loc_gpe_admin1 | 0.000711 |
92
+ | I-nam_pro_title_treaty | 0.000697 |
93
+ | B-nam_eve_human_sport | 0.000684 |
94
+ | I-nam_org_political_party | 0.000680 |
95
+ | B-nam_pro_software | 0.000636 |
96
+ | I-nam_fac_road | 0.000623 |
97
+ | B-nam_adj | 0.000561 |
98
+ | B-nam_loc_gpe_admin3 | 0.000518 |
99
+ | B-nam_pro_model_car | 0.000509 |
100
+ | B-nam_loc_hydronym_river | 0.000491 |
101
+ | B-nam_oth | 0.000452 |
102
+ | B-nam_pro_title_document | 0.000430 |
103
+ | B-nam_loc_astronomical | 0.000425 |
104
+ | B-nam_oth_currency | 0.000421 |
105
+ | B-nam_adj_city | 0.000390 |
106
+ | I-nam_eve | 0.000373 |
107
+ | I-nam_org_group_band | 0.000364 |
108
+ | B-nam_org_group_band | 0.000342 |
109
+ | I-nam_pro_media_web | 0.000329 |
110
+ | I-nam_pro_title_book | 0.000329 |
111
+ | B-nam_loc_gpe_admin2 | 0.000329 |
112
+ | I-nam_pro_software | 0.000320 |
113
+ | I-nam_eve_human_cultural | 0.000298 |
114
+ | I-nam_oth_tech | 0.000294 |
115
+ | B-nam_loc_gpe_district | 0.000294 |
116
+ | I-nam_oth | 0.000285 |
117
+ | B-nam_loc_land_continent | 0.000268 |
118
+ | B-nam_loc_country_region | 0.000268 |
119
+ | B-nam_loc_land_mountain | 0.000241 |
120
+ | I-nam_pro_title_article | 0.000228 |
121
+ | B-nam_pro_title_book | 0.000224 |
122
+ | B-nam_loc_historical_region | 0.000219 |
123
+ | B-nam_loc | 0.000211 |
124
+ | B-nam_eve | 0.000211 |
125
+ | B-nam_org_group | 0.000193 |
126
+ | B-nam_loc_land_island | 0.000193 |
127
+ | I-nam_pro_title_tv | 0.000193 |
128
+ | I-nam_pro_title_album | 0.000189 |
129
+ | B-nam_pro_media_tv | 0.000184 |
130
+ | B-nam_liv_habitant | 0.000184 |
131
+ | B-nam_eve_human_cultural | 0.000184 |
132
+ | I-nam_pro_title_song | 0.000184 |
133
+ | I-nam_oth_license | 0.000180 |
134
+ | B-nam_pro_title_tv | 0.000180 |
135
+ | I-nam_oth_position | 0.000175 |
136
+ | I-nam_loc_country_region | 0.000171 |
137
+ | I-nam_loc_gpe_admin1 | 0.000171 |
138
+ | B-nam_oth_license | 0.000167 |
139
+ | B-nam_num_house | 0.000149 |
140
+ | B-nam_pro_title_treaty | 0.000145 |
141
+ | B-nam_fac_system | 0.000145 |
142
+ | I-nam_loc_gpe_admin3 | 0.000140 |
143
+ | B-nam_loc_gpe_subdivision | 0.000140 |
144
+ | I-nam_eve_human_holiday | 0.000136 |
145
+ | I-nam_org_group | 0.000136 |
146
+ | B-nam_loc_land_region | 0.000132 |
147
+ | I-nam_pro_award | 0.000132 |
148
+ | I-nam_loc_land_mountain | 0.000132 |
149
+ | I-nam_loc_astronomical | 0.000132 |
150
+ | B-nam_pro_title_album | 0.000127 |
151
+ | I-nam_pro_software_game | 0.000123 |
152
+ | B-nam_adj_person | 0.000118 |
153
+ | B-nam_fac_square | 0.000114 |
154
+ | I-nam_pro_media_radio | 0.000114 |
155
+ | B-nam_pro_award | 0.000110 |
156
+ | B-nam_eve_human_holiday | 0.000110 |
157
+ | I-nam_loc | 0.000101 |
158
+ | B-nam_pro_title_song | 0.000096 |
159
+ | I-nam_loc_gpe_subdivision | 0.000096 |
160
+ | B-nam_pro_media_radio | 0.000088 |
161
+ | I-nam_loc_gpe_district | 0.000088 |
162
+ | B-nam_pro_vehicle | 0.000088 |
163
+ | I-nam_loc_land_island | 0.000083 |
164
+ | I-nam_fac_park | 0.000083 |
165
+ | B-nam_oth_position | 0.000083 |
166
+ | B-nam_liv_animal | 0.000083 |
167
+ | I-nam_pro | 0.000083 |
168
+ | B-nam_pro | 0.000079 |
169
+ | I-nam_loc_historical_region | 0.000079 |
170
+ | I-nam_loc_land_region | 0.000075 |
171
+ | I-nam_liv_god | 0.000075 |
172
+ | I-nam_num_phone | 0.000075 |
173
+ | I-nam_fac_bridge | 0.000075 |
174
+ | I-nam_pro_media_tv | 0.000070 |
175
+ | B-nam_oth_www | 0.000070 |
176
+ | B-nam_num_phone | 0.000070 |
177
+ | B-nam_pro_title_article | 0.000066 |
178
+ | B-nam_oth_data_format | 0.000066 |
179
+ | B-nam_fac_bridge | 0.000061 |
180
+ | B-nam_liv_character | 0.000057 |
181
+ | I-nam_org_organization_sub | 0.000053 |
182
+ | B-nam_pro_software_game | 0.000053 |
183
+ | B-nam_loc_hydronym_lake | 0.000053 |
184
+ | B-nam_loc_gpe_conurbation | 0.000053 |
185
+ | B-nam_pro_media | 0.000048 |
186
+ | I-nam_fac_square | 0.000044 |
187
+ | B-nam_loc_land | 0.000044 |
188
+ | B-nam_loc_land_peak | 0.000044 |
189
+ | B-nam_fac_park | 0.000039 |
190
+ | B-nam_org_organization_sub | 0.000035 |
191
+ | I-nam_loc_hydronym_lake | 0.000035 |
192
+ | B-nam_loc_hydronym | 0.000035 |
193
+ | I-nam_pro_vehicle | 0.000035 |
194
+ | I-nam_loc_gpe_conurbation | 0.000035 |
195
+ | I-nam_fac_goe_stop | 0.000035 |
196
+ | I-nam_fac_system | 0.000031 |
197
+ | I-nam_pro_media | 0.000031 |
198
+ | I-nam_loc_gpe_admin2 | 0.000031 |
199
+ | I-nam_loc_land | 0.000026 |
200
+ | B-nam_loc_hydronym_sea | 0.000026 |
201
+ | B-nam_loc_hydronym_ocean | 0.000026 |
202
+ | I-nam_org_nation | 0.000026 |
203
+ | I-nam_liv_character | 0.000022 |
204
+ | I-nam_oth_www | 0.000022 |
205
+ | B-nam_fac_goe_stop | 0.000022 |
206
+ | I-nam_loc_hydronym_sea | 0.000018 |
207
+ | I-nam_oth_currency | 0.000018 |
208
+ | I-nam_loc_hydronym | 0.000018 |
209
+ | I-nam_liv_animal | 0.000018 |
210
+ | I-nam_loc_hydronym_river | 0.000018 |
211
+ | I-nam_oth_data_format | 0.000013 |
212
+ | I-nam_loc_land_continent | 0.000009 |
213
+ | I-nam_loc_land_peak | 0.000009 |
214
+ | I-nam_num_house | 0.000009 |
215
+ | I-nam_loc_hydronym_ocean | 0.000009 |
216
+
217
+ ## Citation
218
+
219
+ ```
220
+ @inproceedings{broda-etal-2012-kpwr,
221
+ title = "{KPW}r: Towards a Free Corpus of {P}olish",
222
+ author = "Broda, Bartosz and
223
+ Marci{\'n}czuk, Micha{\l} and
224
+ Maziarz, Marek and
225
+ Radziszewski, Adam and
226
+ Wardy{\'n}ski, Adam",
227
+ booktitle = "Proceedings of the Eighth International Conference on Language Resources and Evaluation ({LREC}'12)",
228
+ month = may,
229
+ year = "2012",
230
+ address = "Istanbul, Turkey",
231
+ publisher = "European Language Resources Association (ELRA)",
232
+ url = "http://www.lrec-conf.org/proceedings/lrec2012/pdf/965_Paper.pdf",
233
+ pages = "3218--3222",
234
+ abstract = "This paper presents our efforts aimed at collecting and annotating a free Polish corpus. The corpus will serve for us as training and testing material for experiments with Machine Learning algorithms. As others may also benefit from the resource, we are going to release it under a Creative Commons licence, which is hoped to remove unnecessary usage restrictions, but also to facilitate reproduction of our experimental results. The corpus is being annotated with various types of linguistic entities: chunks and named entities, selected syntactic and semantic relations, word senses and anaphora. We report on the current state of the project as well as our ultimate goals.",
235
+ }
236
+ ```
237
+
238
+ ## License
239
+
240
+ ```
241
+ Creative Commons Attribution 3.0 Unported Licence
242
+ ```
243
+
244
+ ## Links
245
+
246
+ [HuggingFace](https://huggingface.co/datasets/clarin-pl/kpwr-ner)
247
+
248
+ [Source](https://clarin-pl.eu/index.php/kpwr-en/)
249
+
250
+ [Paper](https://aclanthology.org/L12-1574/)
251
+
252
+ [KPWr annotation guidelines](http://www.nlp.pwr.wroc.pl/narzedzia-i-zasoby/zasoby/kpwr-lemma/16-narzedzia-zasoby/79-wytyczne)
253
+
254
+ [KPWr annotation guidelines - named entities](https://clarin-pl.eu/dspace/handle/11321/294)
255
+
256
+ ## Examples
257
+
258
+ ### Loading
259
+
260
+ ```python
261
+ from pprint import pprint
262
+
263
+ from datasets import load_dataset
264
+
265
+ dataset = load_dataset("clarin-pl/kpwr-ner")
266
+ pprint(dataset['train'][0])
267
+
268
+ # {'lemmas': ['roborally', 'czy', 'wysoki', 'napięcie', '?'],
269
+ # 'ner': [73, 160, 73, 151, 160],
270
+ # 'orth': ['subst:sg:nom:n',
271
+ # 'qub',
272
+ # 'adj:sg:nom:n:pos',
273
+ # 'subst:sg:nom:n',
274
+ # 'interp'],
275
+ # 'tokens': ['RoboRally', 'czy', 'Wysokie', 'napięcie', '?']}
276
+ ```
277
+
278
+ ### Evaluation
279
+
280
+ ```python
281
+ import random
282
+ from pprint import pprint
283
+
284
+ from datasets import load_dataset, load_metric
285
+
286
+ dataset = load_dataset("clarin-pl/kpwr-ner")
287
+ references = dataset["test"]["ner"]
288
+
289
+ # generate random predictions
290
+ predictions = [
291
+ [
292
+ random.randrange(dataset["train"].features["ner"].feature.num_classes)
293
+ for _ in range(len(labels))
294
+ ]
295
+ for labels in references
296
+ ]
297
+
298
+ # transform to original names of labels
299
+ references_named = [
300
+ [dataset["train"].features["ner"].feature.names[label] for label in labels]
301
+ for labels in references
302
+ ]
303
+ predictions_named = [
304
+ [dataset["train"].features["ner"].feature.names[label] for label in labels]
305
+ for labels in predictions
306
+ ]
307
+
308
+ # utilise seqeval to evaluate
309
+ seqeval = load_metric("seqeval")
310
+ seqeval_score = seqeval.compute(
311
+ predictions=predictions_named, references=references_named, scheme="IOB2"
312
+ )
313
+
314
+ pprint(seqeval_score, depth=1)
315
+
316
+ # {'nam_adj': {...},
317
+ # 'nam_adj_city': {...},
318
+ # 'nam_adj_country': {...},
319
+ # 'nam_adj_person': {...},
320
+ # 'nam_eve': {...},
321
+ # 'nam_eve_human': {...},
322
+ # 'nam_eve_human_cultural': {...},
323
+ # 'nam_eve_human_holiday': {...},
324
+ # 'nam_eve_human_sport': {...},
325
+ # 'nam_fac_bridge': {...},
326
+ # 'nam_fac_goe': {...},
327
+ # 'nam_fac_goe_stop': {...},
328
+ # 'nam_fac_park': {...},
329
+ # 'nam_fac_road': {...},
330
+ # 'nam_fac_square': {...},
331
+ # 'nam_fac_system': {...},
332
+ # 'nam_liv_animal': {...},
333
+ # 'nam_liv_character': {...},
334
+ # 'nam_liv_god': {...},
335
+ # 'nam_liv_habitant': {...},
336
+ # 'nam_liv_person': {...},
337
+ # 'nam_loc': {...},
338
+ # 'nam_loc_astronomical': {...},
339
+ # 'nam_loc_country_region': {...},
340
+ # 'nam_loc_gpe_admin1': {...},
341
+ # 'nam_loc_gpe_admin2': {...},
342
+ # 'nam_loc_gpe_admin3': {...},
343
+ # 'nam_loc_gpe_city': {...},
344
+ # 'nam_loc_gpe_conurbation': {...},
345
+ # 'nam_loc_gpe_country': {...},
346
+ # 'nam_loc_gpe_district': {...},
347
+ # 'nam_loc_gpe_subdivision': {...},
348
+ # 'nam_loc_historical_region': {...},
349
+ # 'nam_loc_hydronym': {...},
350
+ # 'nam_loc_hydronym_lake': {...},
351
+ # 'nam_loc_hydronym_ocean': {...},
352
+ # 'nam_loc_hydronym_river': {...},
353
+ # 'nam_loc_hydronym_sea': {...},
354
+ # 'nam_loc_land': {...},
355
+ # 'nam_loc_land_continent': {...},
356
+ # 'nam_loc_land_island': {...},
357
+ # 'nam_loc_land_mountain': {...},
358
+ # 'nam_loc_land_peak': {...},
359
+ # 'nam_loc_land_region': {...},
360
+ # 'nam_num_house': {...},
361
+ # 'nam_num_phone': {...},
362
+ # 'nam_org_company': {...},
363
+ # 'nam_org_group': {...},
364
+ # 'nam_org_group_band': {...},
365
+ # 'nam_org_group_team': {...},
366
+ # 'nam_org_institution': {...},
367
+ # 'nam_org_nation': {...},
368
+ # 'nam_org_organization': {...},
369
+ # 'nam_org_organization_sub': {...},
370
+ # 'nam_org_political_party': {...},
371
+ # 'nam_oth': {...},
372
+ # 'nam_oth_currency': {...},
373
+ # 'nam_oth_data_format': {...},
374
+ # 'nam_oth_license': {...},
375
+ # 'nam_oth_position': {...},
376
+ # 'nam_oth_tech': {...},
377
+ # 'nam_oth_www': {...},
378
+ # 'nam_pro': {...},
379
+ # 'nam_pro_award': {...},
380
+ # 'nam_pro_brand': {...},
381
+ # 'nam_pro_media': {...},
382
+ # 'nam_pro_media_periodic': {...},
383
+ # 'nam_pro_media_radio': {...},
384
+ # 'nam_pro_media_tv': {...},
385
+ # 'nam_pro_media_web': {...},
386
+ # 'nam_pro_model_car': {...},
387
+ # 'nam_pro_software': {...},
388
+ # 'nam_pro_software_game': {...},
389
+ # 'nam_pro_title': {...},
390
+ # 'nam_pro_title_album': {...},
391
+ # 'nam_pro_title_article': {...},
392
+ # 'nam_pro_title_book': {...},
393
+ # 'nam_pro_title_document': {...},
394
+ # 'nam_pro_title_song': {...},
395
+ # 'nam_pro_title_treaty': {...},
396
+ # 'nam_pro_title_tv': {...},
397
+ # 'nam_pro_vehicle': {...},
398
+ # 'overall_accuracy': 0.006156203762418094,
399
+ # 'overall_f1': 0.0009844258777797407,
400
+ # 'overall_precision': 0.0005213624939842789,
401
+ # 'overall_recall': 0.008803611738148984}
402
+ ```
data/kpwr-ner-n82-test.iob ADDED
The diff for this file is too large to render. See raw diff
 
data/kpwr-ner-n82-train-tune.iob ADDED
The diff for this file is too large to render. See raw diff
 
kpwr-ner.py ADDED
@@ -0,0 +1,266 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # coding=utf-8
2
+ # Copyright 2021 The TensorFlow Datasets Authors and the HuggingFace Datasets Authors.
3
+ #
4
+ # Licensed under the Apache License, Version 2.0 (the "License");
5
+ # you may not use this file except in compliance with the License.
6
+ # You may obtain a copy of the License at
7
+ #
8
+ # http://www.apache.org/licenses/LICENSE-2.0
9
+ #
10
+ # Unless required by applicable law or agreed to in writing, software
11
+ # distributed under the License is distributed on an "AS IS" BASIS,
12
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13
+ # See the License for the specific language governing permissions and
14
+ # limitations under the License.
15
+
16
+ # Lint as: python3
17
+ """KPWR-NER tagging dataset."""
18
+
19
+ import csv
20
+ from typing import List, Tuple, Dict, Generator
21
+
22
+ import datasets
23
+
24
+ _DESCRIPTION = """KPWR-NER tagging dataset."""
25
+
26
+ _URLS = {
27
+ "train": "https://huggingface.co/datasets/clarin-pl/kpwr-ner/resolve/main/data/kpwr-ner-n82-train-tune.iob",
28
+ "test": "https://huggingface.co/datasets/clarin-pl/kpwr-ner/resolve/main/data/kpwr-ner-n82-test.iob",
29
+ }
30
+
31
+ _HOMEPAGE = "https://clarin-pl.eu/dspace/handle/11321/294"
32
+
33
+ _NER_TAGS = [
34
+ "B-nam_adj",
35
+ "B-nam_adj_city",
36
+ "B-nam_adj_country",
37
+ "B-nam_adj_person",
38
+ "B-nam_eve",
39
+ "B-nam_eve_human",
40
+ "B-nam_eve_human_cultural",
41
+ "B-nam_eve_human_holiday",
42
+ "B-nam_eve_human_sport",
43
+ "B-nam_fac_bridge",
44
+ "B-nam_fac_goe",
45
+ "B-nam_fac_goe_stop",
46
+ "B-nam_fac_park",
47
+ "B-nam_fac_road",
48
+ "B-nam_fac_square",
49
+ "B-nam_fac_system",
50
+ "B-nam_liv_animal",
51
+ "B-nam_liv_character",
52
+ "B-nam_liv_god",
53
+ "B-nam_liv_habitant",
54
+ "B-nam_liv_person",
55
+ "B-nam_loc",
56
+ "B-nam_loc_astronomical",
57
+ "B-nam_loc_country_region",
58
+ "B-nam_loc_gpe_admin1",
59
+ "B-nam_loc_gpe_admin2",
60
+ "B-nam_loc_gpe_admin3",
61
+ "B-nam_loc_gpe_city",
62
+ "B-nam_loc_gpe_conurbation",
63
+ "B-nam_loc_gpe_country",
64
+ "B-nam_loc_gpe_district",
65
+ "B-nam_loc_gpe_subdivision",
66
+ "B-nam_loc_historical_region",
67
+ "B-nam_loc_hydronym",
68
+ "B-nam_loc_hydronym_lake",
69
+ "B-nam_loc_hydronym_ocean",
70
+ "B-nam_loc_hydronym_river",
71
+ "B-nam_loc_hydronym_sea",
72
+ "B-nam_loc_land",
73
+ "B-nam_loc_land_continent",
74
+ "B-nam_loc_land_island",
75
+ "B-nam_loc_land_mountain",
76
+ "B-nam_loc_land_peak",
77
+ "B-nam_loc_land_region",
78
+ "B-nam_num_house",
79
+ "B-nam_num_phone",
80
+ "B-nam_org_company",
81
+ "B-nam_org_group",
82
+ "B-nam_org_group_band",
83
+ "B-nam_org_group_team",
84
+ "B-nam_org_institution",
85
+ "B-nam_org_nation",
86
+ "B-nam_org_organization",
87
+ "B-nam_org_organization_sub",
88
+ "B-nam_org_political_party",
89
+ "B-nam_oth",
90
+ "B-nam_oth_currency",
91
+ "B-nam_oth_data_format",
92
+ "B-nam_oth_license",
93
+ "B-nam_oth_position",
94
+ "B-nam_oth_tech",
95
+ "B-nam_oth_www",
96
+ "B-nam_pro",
97
+ "B-nam_pro_award",
98
+ "B-nam_pro_brand",
99
+ "B-nam_pro_media",
100
+ "B-nam_pro_media_periodic",
101
+ "B-nam_pro_media_radio",
102
+ "B-nam_pro_media_tv",
103
+ "B-nam_pro_media_web",
104
+ "B-nam_pro_model_car",
105
+ "B-nam_pro_software",
106
+ "B-nam_pro_software_game",
107
+ "B-nam_pro_title",
108
+ "B-nam_pro_title_album",
109
+ "B-nam_pro_title_article",
110
+ "B-nam_pro_title_book",
111
+ "B-nam_pro_title_document",
112
+ "B-nam_pro_title_song",
113
+ "B-nam_pro_title_treaty",
114
+ "B-nam_pro_title_tv",
115
+ "B-nam_pro_vehicle",
116
+ "I-nam_adj_country",
117
+ "I-nam_eve",
118
+ "I-nam_eve_human",
119
+ "I-nam_eve_human_cultural",
120
+ "I-nam_eve_human_holiday",
121
+ "I-nam_eve_human_sport",
122
+ "I-nam_fac_bridge",
123
+ "I-nam_fac_goe",
124
+ "I-nam_fac_goe_stop",
125
+ "I-nam_fac_park",
126
+ "I-nam_fac_road",
127
+ "I-nam_fac_square",
128
+ "I-nam_fac_system",
129
+ "I-nam_liv_animal",
130
+ "I-nam_liv_character",
131
+ "I-nam_liv_god",
132
+ "I-nam_liv_person",
133
+ "I-nam_loc",
134
+ "I-nam_loc_astronomical",
135
+ "I-nam_loc_country_region",
136
+ "I-nam_loc_gpe_admin1",
137
+ "I-nam_loc_gpe_admin2",
138
+ "I-nam_loc_gpe_admin3",
139
+ "I-nam_loc_gpe_city",
140
+ "I-nam_loc_gpe_conurbation",
141
+ "I-nam_loc_gpe_country",
142
+ "I-nam_loc_gpe_district",
143
+ "I-nam_loc_gpe_subdivision",
144
+ "I-nam_loc_historical_region",
145
+ "I-nam_loc_hydronym",
146
+ "I-nam_loc_hydronym_lake",
147
+ "I-nam_loc_hydronym_ocean",
148
+ "I-nam_loc_hydronym_river",
149
+ "I-nam_loc_hydronym_sea",
150
+ "I-nam_loc_land",
151
+ "I-nam_loc_land_continent",
152
+ "I-nam_loc_land_island",
153
+ "I-nam_loc_land_mountain",
154
+ "I-nam_loc_land_peak",
155
+ "I-nam_loc_land_region",
156
+ "I-nam_num_house",
157
+ "I-nam_num_phone",
158
+ "I-nam_org_company",
159
+ "I-nam_org_group",
160
+ "I-nam_org_group_band",
161
+ "I-nam_org_group_team",
162
+ "I-nam_org_institution",
163
+ "I-nam_org_nation",
164
+ "I-nam_org_organization",
165
+ "I-nam_org_organization_sub",
166
+ "I-nam_org_political_party",
167
+ "I-nam_oth",
168
+ "I-nam_oth_currency",
169
+ "I-nam_oth_data_format",
170
+ "I-nam_oth_license",
171
+ "I-nam_oth_position",
172
+ "I-nam_oth_tech",
173
+ "I-nam_oth_www",
174
+ "I-nam_pro",
175
+ "I-nam_pro_award",
176
+ "I-nam_pro_brand",
177
+ "I-nam_pro_media",
178
+ "I-nam_pro_media_periodic",
179
+ "I-nam_pro_media_radio",
180
+ "I-nam_pro_media_tv",
181
+ "I-nam_pro_media_web",
182
+ "I-nam_pro_model_car",
183
+ "I-nam_pro_software",
184
+ "I-nam_pro_software_game",
185
+ "I-nam_pro_title",
186
+ "I-nam_pro_title_album",
187
+ "I-nam_pro_title_article",
188
+ "I-nam_pro_title_book",
189
+ "I-nam_pro_title_document",
190
+ "I-nam_pro_title_song",
191
+ "I-nam_pro_title_treaty",
192
+ "I-nam_pro_title_tv",
193
+ "I-nam_pro_vehicle",
194
+ "O",
195
+ ]
196
+
197
+
198
+ class KPWRNER(datasets.GeneratorBasedBuilder):
199
+ def _info(self) -> datasets.DatasetInfo:
200
+ return datasets.DatasetInfo(
201
+ description=_DESCRIPTION,
202
+ features=datasets.Features(
203
+ {
204
+ "tokens": datasets.Sequence(datasets.Value("string")),
205
+ "lemmas": datasets.Sequence(datasets.Value("string")),
206
+ "orth": datasets.Sequence(datasets.Value("string")),
207
+ "ner": datasets.Sequence(
208
+ datasets.features.ClassLabel(
209
+ names=_NER_TAGS, num_classes=len(_NER_TAGS)
210
+ )
211
+ ),
212
+ }
213
+ ),
214
+ homepage=_HOMEPAGE,
215
+ )
216
+
217
+ def _split_generators(
218
+ self, dl_manager: datasets.DownloadManager
219
+ ) -> List[datasets.SplitGenerator]:
220
+ urls_to_download = _URLS
221
+ downloaded_files = dl_manager.download_and_extract(urls_to_download)
222
+ return [
223
+ datasets.SplitGenerator(
224
+ name=datasets.Split.TRAIN,
225
+ gen_kwargs={"filepath": downloaded_files["train"]},
226
+ ),
227
+ datasets.SplitGenerator(
228
+ name=datasets.Split.TEST,
229
+ gen_kwargs={"filepath": downloaded_files["test"]},
230
+ ),
231
+ ]
232
+
233
+ def _generate_examples(
234
+ self, filepath: str
235
+ ) -> Generator[Tuple[int, Dict[str, str]], None, None]:
236
+ with open(filepath, "r", encoding="utf-8") as f:
237
+ reader = csv.reader(f, delimiter="\t", quoting=csv.QUOTE_NONE)
238
+
239
+ tokens = []
240
+ lemma = []
241
+ orth = []
242
+ ner = []
243
+ gid = 0
244
+
245
+ for line in reader:
246
+ if not line:
247
+ yield gid, {
248
+ "tokens": tokens,
249
+ "lemmas": lemma,
250
+ "orth": orth,
251
+ "ner": ner,
252
+ }
253
+ gid += 1
254
+ tokens = []
255
+ lemma = []
256
+ orth = []
257
+ ner = []
258
+
259
+ elif len(line) == 1: # ignore DOCS
260
+ continue
261
+
262
+ else:
263
+ tokens.append(line[0])
264
+ lemma.append(line[1])
265
+ orth.append(line[2])
266
+ ner.append(line[3])