qanastek commited on
Commit
fbe630d
1 Parent(s): 324c958

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +683 -683
README.md CHANGED
@@ -1,683 +1,683 @@
1
- ---
2
- annotations_creators:
3
- - machine-generated
4
- - expert-generated
5
- language_creators:
6
- - found
7
- languages:
8
- - af-ZA
9
- - am-ET
10
- - ar-SA
11
- - az-AZ
12
- - bn-BD
13
- - cy-GB
14
- - da-DK
15
- - de-DE
16
- - el-GR
17
- - en-US
18
- - es-ES
19
- - fa-IR
20
- - fi-FI
21
- - fr-FR
22
- - he-IL
23
- - hi-IN
24
- - hu-HU
25
- - hy-AM
26
- - id-ID
27
- - is-IS
28
- - it-IT
29
- - ja-JP
30
- - jv-ID
31
- - ka-GE
32
- - km-KH
33
- - kn-IN
34
- - ko-KR
35
- - lv-LV
36
- - ml-IN
37
- - mn-MN
38
- - ms-MY
39
- - my-MM
40
- - nb-NO
41
- - nl-NL
42
- - pl-PL
43
- - pt-PT
44
- - ro-RO
45
- - ru-RU
46
- - sl-SL
47
- - sq-AL
48
- - sv-SE
49
- - sw-KE
50
- - ta-IN
51
- - te-IN
52
- - th-TH
53
- - tl-PH
54
- - tr-TR
55
- - ur-PK
56
- - vi-VN
57
- - zh-CN
58
- - zh-TW
59
- licenses:
60
- - Copyright Amazon.com Inc. or its affiliates.
61
- multilinguality:
62
- - af-ZA
63
- - am-ET
64
- - ar-SA
65
- - az-AZ
66
- - bn-BD
67
- - cy-GB
68
- - da-DK
69
- - de-DE
70
- - el-GR
71
- - en-US
72
- - es-ES
73
- - fa-IR
74
- - fi-FI
75
- - fr-FR
76
- - he-IL
77
- - hi-IN
78
- - hu-HU
79
- - hy-AM
80
- - id-ID
81
- - is-IS
82
- - it-IT
83
- - ja-JP
84
- - jv-ID
85
- - ka-GE
86
- - km-KH
87
- - kn-IN
88
- - ko-KR
89
- - lv-LV
90
- - ml-IN
91
- - mn-MN
92
- - ms-MY
93
- - my-MM
94
- - nb-NO
95
- - nl-NL
96
- - pl-PL
97
- - pt-PT
98
- - ro-RO
99
- - ru-RU
100
- - sl-SL
101
- - sq-AL
102
- - sv-SE
103
- - sw-KE
104
- - ta-IN
105
- - te-IN
106
- - th-TH
107
- - tl-PH
108
- - tr-TR
109
- - ur-PK
110
- - vi-VN
111
- - zh-CN
112
- - zh-TW
113
- pretty_name: MASSIVE
114
- size_categories:
115
- - 100K<n<1M
116
- source_datasets:
117
- - original
118
- task_categories:
119
- - text-classification
120
- task_ids:
121
- - intent-classification
122
- - multi-class-classification
123
- - natural-language-understanding
124
- ---
125
-
126
- # MASSIVE: A 1M-Example Multilingual Natural Language Understanding Dataset with 51 Typologically-Diverse Languages
127
-
128
- ## Table of Contents
129
- - [Dataset Card for [Needs More Information]](#dataset-card-for-needs-more-information)
130
- - [Table of Contents](#table-of-contents)
131
- - [Dataset Description](#dataset-description)
132
- - [Dataset Summary](#dataset-summary)
133
- - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards)
134
- - [Languages](#languages)
135
- - [Dataset Structure](#dataset-structure)
136
- - [Data Instances](#data-instances)
137
- - [Data Fields](#data-fields)
138
- - [Data Splits](#data-splits)
139
- - [Dataset Creation](#dataset-creation)
140
- - [Curation Rationale](#curation-rationale)
141
- - [Source Data](#source-data)
142
- - [Initial Data Collection and Normalization](#initial-data-collection-and-normalization)
143
- - [Who are the source language producers?](#who-are-the-source-language-producers)
144
- - [Personal and Sensitive Information](#personal-and-sensitive-information)
145
- - [Considerations for Using the Data](#considerations-for-using-the-data)
146
- - [Other Known Limitations](#other-known-limitations)
147
- - [Additional Information](#additional-information)
148
- - [Dataset Curators](#dataset-curators)
149
- - [Licensing Information](#licensing-information)
150
- - [No Warranty](#no-warranty)
151
- - [Citation Information](#citation-information)
152
-
153
- ## Dataset Description
154
-
155
- - **Homepage:** https://github.com/alexa/massive
156
- - **Repository:** https://github.com/alexa/massive
157
- - **Paper:** https://arxiv.org/abs/2204.08582
158
- - **Leaderboard:** https://eval.ai/web/challenges/challenge-page/1697/overview
159
- - **Point of Contact:** [GitHub](https://github.com/alexa/massive/issues)
160
-
161
- ### Dataset Summary
162
-
163
- MASSIVE is a parallel dataset of > 1M utterances across 51 languages with annotations for the Natural Language Understanding tasks of intent prediction and slot annotation. Utterances span 60 intents and include 55 slot types. MASSIVE was created by localizing the SLURP dataset, composed of general Intelligent Voice Assistant single-shot interactions.
164
-
165
- | Name | # Lang | Utt/Lang | Domains | Intents | Slots |
166
- |:-------------------------------------------------------------------------------:|:-------:|:--------------:|:-------:|:--------:|:------:|
167
- | MASSIVE | 51 | 19,521 | 18 | 60 | 55 |
168
- | SLURP (Bastianelli et al., 2020) | 1 | 16,521 | 18 | 60 | 55 |
169
- | NLU Evaluation Data (Liu et al., 2019) | 1 | 25,716 | 18 | 54 | 56 |
170
- | Airline Travel Information System (ATIS) (Price, 1990) | 1 | 5,871 | 1 | 26 | 129 |
171
- | ATIS with Hindi and Turkish (Upadhyay et al., 2018) | 3 | 1,315-5,871 | 1 | 26 | 129 |
172
- | MultiATIS++ (Xu et al., 2020) | 9 | 1,422-5,897 | 1 | 21-26 | 99-140 |
173
- | Snips (Coucke et al., 2018) | 1 | 14,484 | - | 7 | 53 |
174
- | Snips with French (Saade et al., 2019) | 2 | 4,818 | 2 | 14-15 | 11-12 |
175
- | Task Oriented Parsing (TOP) (Gupta et al., 2018) | 1 | 44,873 | 2 | 25 | 36 |
176
- | Multilingual Task-Oriented Semantic Parsing (MTOP) (Li et al., 2021) | 6 | 15,195-22,288 | 11 | 104-113 | 72-75 |
177
- | Cross-Lingual Multilingual Task Oriented Dialog (Schuster et al., 2019) | 3 | 5,083-43,323 | 3 | 12 | 11 |
178
- | Microsoft Dialog Challenge (Li et al., 2018) | 1 | 38,276 | 3 | 11 | 29 |
179
- | Fluent Speech Commands (FSC) (Lugosch et al., 2019) | 1 | 30,043 | - | 31 | - |
180
- | Chinese Audio-Textual Spoken Language Understanding (CATSLU) (Zhu et al., 2019) | 1 | 16,258 | 4 | - | 94 |
181
-
182
- ### Supported Tasks and Leaderboards
183
-
184
- The dataset can be used to train a model for `natural-language-understanding` (NLU) :
185
-
186
- - `intent-classification`
187
- - `multi-class-classification`
188
- - `natural-language-understanding`
189
-
190
- ### Languages
191
-
192
- The corpora consists of parallel sentences from 51 languages :
193
-
194
- - `Afrikaans - South Africa (af-ZA)`
195
- - `Amharic - Ethiopia (am-ET)`
196
- - `Arabic - Saudi Arabia (ar-SA)`
197
- - `Azeri - Azerbaijan (az-AZ)`
198
- - `Bengali - Bangladesh (bn-BD)`
199
- - `Chinese - China (zh-CN)`
200
- - `Chinese - Taiwan (zh-TW)`
201
- - `Danish - Denmark (da-DK)`
202
- - `German - Germany (de-DE)`
203
- - `Greek - Greece (el-GR)`
204
- - `English - United States (en-US)`
205
- - `Spanish - Spain (es-ES)`
206
- - `Farsi - Iran (fa-IR)`
207
- - `Finnish - Finland (fi-FI)`
208
- - `French - France (fr-FR)`
209
- - `Hebrew - Israel (he-IL)`
210
- - `Hungarian - Hungary (hu-HU)`
211
- - `Armenian - Armenia (hy-AM)`
212
- - `Indonesian - Indonesia (id-ID)`
213
- - `Icelandic - Iceland (is-IS)`
214
- - `Italian - Italy (it-IT)`
215
- - `Japanese - Japan (ja-JP)`
216
- - `Javanese - Indonesia (jv-ID)`
217
- - `Georgian - Georgia (ka-GE)`
218
- - `Khmer - Cambodia (km-KH)`
219
- - `Korean - Korea (ko-KR)`
220
- - `Latvian - Latvia (lv-LV)`
221
- - `Mongolian - Mongolia (mn-MN)`
222
- - `Malay - Malaysia (ms-MY)`
223
- - `Burmese - Myanmar (my-MM)`
224
- - `Norwegian - Norway (nb-NO)`
225
- - `Dutch - Netherlands (nl-NL)`
226
- - `Polish - Poland (pl-PL)`
227
- - `Portuguese - Portugal (pt-PT)`
228
- - `Romanian - Romania (ro-RO)`
229
- - `Russian - Russia (ru-RU)`
230
- - `Slovanian - Slovania (sl-SL)`
231
- - `Albanian - Albania (sq-AL)`
232
- - `Swedish - Sweden (sv-SE)`
233
- - `Swahili - Kenya (sw-KE)`
234
- - `Hindi - India (hi-IN)`
235
- - `Kannada - India (kn-IN)`
236
- - `Malayalam - India (ml-IN)`
237
- - `Tamil - India (ta-IN)`
238
- - `Telugu - India (te-IN)`
239
- - `Thai - Thailand (th-TH)`
240
- - `Tagalog - Philippines (tl-PH)`
241
- - `Turkish - Turkey (tr-TR)`
242
- - `Urdu - Pakistan (ur-PK)`
243
- - `Vietnamese - Vietnam (vi-VN)`
244
- - `Welsh - United Kingdom (cy-GB)`
245
-
246
- ## Load the dataset with HuggingFace
247
-
248
- ```python
249
- from datasets import load_dataset
250
- dataset = load_dataset("qanastek/MASSIVE", "en-US", split='train')
251
- print(dataset)
252
- print(dataset[0])
253
- ```
254
-
255
- ## Dataset Structure
256
-
257
- ### Data Instances (taken from Alexa Github)
258
-
259
- ```json
260
- {
261
- "id": "0",
262
- "locale": "de-DE",
263
- "partition": "test",
264
- "scenario": "alarm",
265
- "intent": "alarm_set",
266
- "utt": "weck mich diese woche um fünf uhr morgens auf",
267
- "annot_utt": "weck mich [date : diese woche] um [time : fünf uhr morgens] auf",
268
- "worker_id": "8",
269
- "slot_method": [
270
- {
271
- "slot": "time",
272
- "method": "translation"
273
- },
274
- {
275
- "slot": "date",
276
- "method": "translation"
277
- }
278
- ],
279
- "judgments": [
280
- {
281
- "worker_id": "32",
282
- "intent_score": 1,
283
- "slots_score": 0,
284
- "grammar_score": 4,
285
- "spelling_score": 2,
286
- "language_identification": "target"
287
- },
288
- {
289
- "worker_id": "8",
290
- "intent_score": 1,
291
- "slots_score": 1,
292
- "grammar_score": 4,
293
- "spelling_score": 2,
294
- "language_identification": "target"
295
- },
296
- {
297
- "worker_id": "28",
298
- "intent_score": 1,
299
- "slots_score": 1,
300
- "grammar_score": 4,
301
- "spelling_score": 2,
302
- "language_identification": "target"
303
- }
304
- ]
305
- }
306
- ```
307
-
308
- ### Data Fields (taken from Alexa Github)
309
-
310
- `id`: maps to the original ID in the [SLURP](https://github.com/pswietojanski/slurp) collection. Mapping back to the SLURP en-US utterance, this utterance served as the basis for this localization.
311
-
312
- `locale`: is the language and country code accoring to ISO-639-1 and ISO-3166.
313
-
314
- `partition`: is either `train`, `dev`, or `test`, according to the original split in [SLURP](https://github.com/pswietojanski/slurp).
315
-
316
- `scenario`: is the general domain, aka "scenario" in SLURP terminology, of an utterance
317
-
318
- `intent`: is the specific intent of an utterance within a domain formatted as `{scenario}_{intent}`
319
-
320
- `utt`: the raw utterance text without annotations
321
-
322
- `annot_utt`: the text from `utt` with slot annotations formatted as `[{label} : {entity}]`
323
-
324
- `worker_id`: The obfuscated worker ID from MTurk of the worker completing the localization of the utterance. Worker IDs are specific to a locale and do *not* map across locales.
325
-
326
- `slot_method`: for each slot in the utterance, whether that slot was a `translation` (i.e., same expression just in the target language), `localization` (i.e., not the same expression but a different expression was chosen more suitable to the phrase in that locale), or `unchanged` (i.e., the original en-US slot value was copied over without modification).
327
-
328
- `judgments`: Each judgment collected for the localized utterance has 6 keys. `worker_id` is the obfuscated worker ID from MTurk of the worker completing the judgment. Worker IDs are specific to a locale and do *not* map across locales, but *are* consistent across the localization tasks and the judgment tasks, e.g., judgment worker ID 32 in the example above may appear as the localization worker ID for the localization of a different de-DE utterance, in which case it would be the same worker.
329
-
330
- ```plain
331
- intent_score : "Does the sentence match the intent?"
332
- 0: No
333
- 1: Yes
334
- 2: It is a reasonable interpretation of the goal
335
-
336
- slots_score : "Do all these terms match the categories in square brackets?"
337
- 0: No
338
- 1: Yes
339
- 2: There are no words in square brackets (utterance without a slot)
340
-
341
- grammar_score : "Read the sentence out loud. Ignore any spelling, punctuation, or capitalization errors. Does it sound natural?"
342
- 0: Completely unnatural (nonsensical, cannot be understood at all)
343
- 1: Severe errors (the meaning cannot be understood and doesn't sound natural in your language)
344
- 2: Some errors (the meaning can be understood but it doesn't sound natural in your language)
345
- 3: Good enough (easily understood and sounds almost natural in your language)
346
- 4: Perfect (sounds natural in your language)
347
-
348
- spelling_score : "Are all words spelled correctly? Ignore any spelling variances that may be due to differences in dialect. Missing spaces should be marked as a spelling error."
349
- 0: There are more than 2 spelling errors
350
- 1: There are 1-2 spelling errors
351
- 2: All words are spelled correctly
352
-
353
- language_identification : "The following sentence contains words in the following languages (check all that apply)"
354
- 1: target
355
- 2: english
356
- 3: other
357
- 4: target & english
358
- 5: target & other
359
- 6: english & other
360
- 7: target & english & other
361
- ```
362
-
363
- ### Data Splits
364
-
365
- |Language|Train|Dev|Test|
366
- |:---:|:---:|:---:|:---:|
367
- |af-ZA|11514|2033|2974|
368
- |am-ET|11514|2033|2974|
369
- |ar-SA|11514|2033|2974|
370
- |az-AZ|11514|2033|2974|
371
- |bn-BD|11514|2033|2974|
372
- |cy-GB|11514|2033|2974|
373
- |da-DK|11514|2033|2974|
374
- |de-DE|11514|2033|2974|
375
- |el-GR|11514|2033|2974|
376
- |en-US|11514|2033|2974|
377
- |es-ES|11514|2033|2974|
378
- |fa-IR|11514|2033|2974|
379
- |fi-FI|11514|2033|2974|
380
- |fr-FR|11514|2033|2974|
381
- |he-IL|11514|2033|2974|
382
- |hi-IN|11514|2033|2974|
383
- |hu-HU|11514|2033|2974|
384
- |hy-AM|11514|2033|2974|
385
- |id-ID|11514|2033|2974|
386
- |is-IS|11514|2033|2974|
387
- |it-IT|11514|2033|2974|
388
- |ja-JP|11514|2033|2974|
389
- |jv-ID|11514|2033|2974|
390
- |ka-GE|11514|2033|2974|
391
- |km-KH|11514|2033|2974|
392
- |kn-IN|11514|2033|2974|
393
- |ko-KR|11514|2033|2974|
394
- |lv-LV|11514|2033|2974|
395
- |ml-IN|11514|2033|2974|
396
- |mn-MN|11514|2033|2974|
397
- |ms-MY|11514|2033|2974|
398
- |my-MM|11514|2033|2974|
399
- |nb-NO|11514|2033|2974|
400
- |nl-NL|11514|2033|2974|
401
- |pl-PL|11514|2033|2974|
402
- |pt-PT|11514|2033|2974|
403
- |ro-RO|11514|2033|2974|
404
- |ru-RU|11514|2033|2974|
405
- |sl-SL|11514|2033|2974|
406
- |sq-AL|11514|2033|2974|
407
- |sv-SE|11514|2033|2974|
408
- |sw-KE|11514|2033|2974|
409
- |ta-IN|11514|2033|2974|
410
- |te-IN|11514|2033|2974|
411
- |th-TH|11514|2033|2974|
412
- |tl-PH|11514|2033|2974|
413
- |tr-TR|11514|2033|2974|
414
- |ur-PK|11514|2033|2974|
415
- |vi-VN|11514|2033|2974|
416
- |zh-CN|11514|2033|2974|
417
- |zh-TW|11514|2033|2974|
418
-
419
- ## Dataset Creation
420
-
421
- ### Source Data
422
-
423
- #### Who are the source language producers?
424
-
425
- The corpus has been produced and uploaded by Amazon Alexa.
426
-
427
- ### Personal and Sensitive Information
428
-
429
- The corpora is free of personal or sensitive information.
430
-
431
- ## Additional Information
432
-
433
- ### Dataset Curators
434
-
435
- __Hugging Face__: Labrak Yanis (Not affiliated with the original corpus)
436
-
437
- __MASSIVE__: Jack FitzGerald and Christopher Hench and Charith Peris and Scott Mackie and Kay Rottmann and Ana Sanchez and Aaron Nash and Liam Urbach and Vishesh Kakarala and Richa Singh and Swetha Ranganath and Laurie Crist and Misha Britan and Wouter Leeuwis and Gokhan Tur and Prem Natarajan.
438
-
439
- __SLURP__: Bastianelli, Emanuele and Vanzo, Andrea and Swietojanski, Pawel and Rieser, Verena.
440
-
441
- ### Licensing Information
442
-
443
- ```plain
444
- Copyright Amazon.com Inc. or its affiliates.
445
-
446
- Copyright and license details for the data and modified code can be found in NOTICE.md.
447
-
448
- License for massive repo and code, Apache 2.0:
449
-
450
- Apache License
451
- Version 2.0, January 2004
452
- http://www.apache.org/licenses/
453
-
454
- TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
455
-
456
- 1. Definitions.
457
-
458
- "License" shall mean the terms and conditions for use, reproduction,
459
- and distribution as defined by Sections 1 through 9 of this document.
460
-
461
- "Licensor" shall mean the copyright owner or entity authorized by
462
- the copyright owner that is granting the License.
463
-
464
- "Legal Entity" shall mean the union of the acting entity and all
465
- other entities that control, are controlled by, or are under common
466
- control with that entity. For the purposes of this definition,
467
- "control" means (i) the power, direct or indirect, to cause the
468
- direction or management of such entity, whether by contract or
469
- otherwise, or (ii) ownership of fifty percent (50%) or more of the
470
- outstanding shares, or (iii) beneficial ownership of such entity.
471
-
472
- "You" (or "Your") shall mean an individual or Legal Entity
473
- exercising permissions granted by this License.
474
-
475
- "Source" form shall mean the preferred form for making modifications,
476
- including but not limited to software source code, documentation
477
- source, and configuration files.
478
-
479
- "Object" form shall mean any form resulting from mechanical
480
- transformation or translation of a Source form, including but
481
- not limited to compiled object code, generated documentation,
482
- and conversions to other media types.
483
-
484
- "Work" shall mean the work of authorship, whether in Source or
485
- Object form, made available under the License, as indicated by a
486
- copyright notice that is included in or attached to the work
487
- (an example is provided in the Appendix below).
488
-
489
- "Derivative Works" shall mean any work, whether in Source or Object
490
- form, that is based on (or derived from) the Work and for which the
491
- editorial revisions, annotations, elaborations, or other modifications
492
- represent, as a whole, an original work of authorship. For the purposes
493
- of this License, Derivative Works shall not include works that remain
494
- separable from, or merely link (or bind by name) to the interfaces of,
495
- the Work and Derivative Works thereof.
496
-
497
- "Contribution" shall mean any work of authorship, including
498
- the original version of the Work and any modifications or additions
499
- to that Work or Derivative Works thereof, that is intentionally
500
- submitted to Licensor for inclusion in the Work by the copyright owner
501
- or by an individual or Legal Entity authorized to submit on behalf of
502
- the copyright owner. For the purposes of this definition, "submitted"
503
- means any form of electronic, verbal, or written communication sent
504
- to the Licensor or its representatives, including but not limited to
505
- communication on electronic mailing lists, source code control systems,
506
- and issue tracking systems that are managed by, or on behalf of, the
507
- Licensor for the purpose of discussing and improving the Work, but
508
- excluding communication that is conspicuously marked or otherwise
509
- designated in writing by the copyright owner as "Not a Contribution."
510
-
511
- "Contributor" shall mean Licensor and any individual or Legal Entity
512
- on behalf of whom a Contribution has been received by Licensor and
513
- subsequently incorporated within the Work.
514
-
515
- 2. Grant of Copyright License. Subject to the terms and conditions of
516
- this License, each Contributor hereby grants to You a perpetual,
517
- worldwide, non-exclusive, no-charge, royalty-free, irrevocable
518
- copyright license to reproduce, prepare Derivative Works of,
519
- publicly display, publicly perform, sublicense, and distribute the
520
- Work and such Derivative Works in Source or Object form.
521
-
522
- 3. Grant of Patent License. Subject to the terms and conditions of
523
- this License, each Contributor hereby grants to You a perpetual,
524
- worldwide, non-exclusive, no-charge, royalty-free, irrevocable
525
- (except as stated in this section) patent license to make, have made,
526
- use, offer to sell, sell, import, and otherwise transfer the Work,
527
- where such license applies only to those patent claims licensable
528
- by such Contributor that are necessarily infringed by their
529
- Contribution(s) alone or by combination of their Contribution(s)
530
- with the Work to which such Contribution(s) was submitted. If You
531
- institute patent litigation against any entity (including a
532
- cross-claim or counterclaim in a lawsuit) alleging that the Work
533
- or a Contribution incorporated within the Work constitutes direct
534
- or contributory patent infringement, then any patent licenses
535
- granted to You under this License for that Work shall terminate
536
- as of the date such litigation is filed.
537
-
538
- 4. Redistribution. You may reproduce and distribute copies of the
539
- Work or Derivative Works thereof in any medium, with or without
540
- modifications, and in Source or Object form, provided that You
541
- meet the following conditions:
542
-
543
- (a) You must give any other recipients of the Work or
544
- Derivative Works a copy of this License; and
545
-
546
- (b) You must cause any modified files to carry prominent notices
547
- stating that You changed the files; and
548
-
549
- (c) You must retain, in the Source form of any Derivative Works
550
- that You distribute, all copyright, patent, trademark, and
551
- attribution notices from the Source form of the Work,
552
- excluding those notices that do not pertain to any part of
553
- the Derivative Works; and
554
-
555
- (d) If the Work includes a "NOTICE" text file as part of its
556
- distribution, then any Derivative Works that You distribute must
557
- include a readable copy of the attribution notices contained
558
- within such NOTICE file, excluding those notices that do not
559
- pertain to any part of the Derivative Works, in at least one
560
- of the following places: within a NOTICE text file distributed
561
- as part of the Derivative Works; within the Source form or
562
- documentation, if provided along with the Derivative Works; or,
563
- within a display generated by the Derivative Works, if and
564
- wherever such third-party notices normally appear. The contents
565
- of the NOTICE file are for informational purposes only and
566
- do not modify the License. You may add Your own attribution
567
- notices within Derivative Works that You distribute, alongside
568
- or as an addendum to the NOTICE text from the Work, provided
569
- that such additional attribution notices cannot be construed
570
- as modifying the License.
571
-
572
- You may add Your own copyright statement to Your modifications and
573
- may provide additional or different license terms and conditions
574
- for use, reproduction, or distribution of Your modifications, or
575
- for any such Derivative Works as a whole, provided Your use,
576
- reproduction, and distribution of the Work otherwise complies with
577
- the conditions stated in this License.
578
-
579
- 5. Submission of Contributions. Unless You explicitly state otherwise,
580
- any Contribution intentionally submitted for inclusion in the Work
581
- by You to the Licensor shall be under the terms and conditions of
582
- this License, without any additional terms or conditions.
583
- Notwithstanding the above, nothing herein shall supersede or modify
584
- the terms of any separate license agreement you may have executed
585
- with Licensor regarding such Contributions.
586
-
587
- 6. Trademarks. This License does not grant permission to use the trade
588
- names, trademarks, service marks, or product names of the Licensor,
589
- except as required for reasonable and customary use in describing the
590
- origin of the Work and reproducing the content of the NOTICE file.
591
-
592
- 7. Disclaimer of Warranty. Unless required by applicable law or
593
- agreed to in writing, Licensor provides the Work (and each
594
- Contributor provides its Contributions) on an "AS IS" BASIS,
595
- WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
596
- implied, including, without limitation, any warranties or conditions
597
- of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
598
- PARTICULAR PURPOSE. You are solely responsible for determining the
599
- appropriateness of using or redistributing the Work and assume any
600
- risks associated with Your exercise of permissions under this License.
601
-
602
- 8. Limitation of Liability. In no event and under no legal theory,
603
- whether in tort (including negligence), contract, or otherwise,
604
- unless required by applicable law (such as deliberate and grossly
605
- negligent acts) or agreed to in writing, shall any Contributor be
606
- liable to You for damages, including any direct, indirect, special,
607
- incidental, or consequential damages of any character arising as a
608
- result of this License or out of the use or inability to use the
609
- Work (including but not limited to damages for loss of goodwill,
610
- work stoppage, computer failure or malfunction, or any and all
611
- other commercial damages or losses), even if such Contributor
612
- has been advised of the possibility of such damages.
613
-
614
- 9. Accepting Warranty or Additional Liability. While redistributing
615
- the Work or Derivative Works thereof, You may choose to offer,
616
- and charge a fee for, acceptance of support, warranty, indemnity,
617
- or other liability obligations and/or rights consistent with this
618
- License. However, in accepting such obligations, You may act only
619
- on Your own behalf and on Your sole responsibility, not on behalf
620
- of any other Contributor, and only if You agree to indemnify,
621
- defend, and hold each Contributor harmless for any liability
622
- incurred by, or claims asserted against, such Contributor by reason
623
- of your accepting any such warranty or additional liability.
624
-
625
- END OF TERMS AND CONDITIONS
626
-
627
- APPENDIX: How to apply the Apache License to your work.
628
-
629
- To apply the Apache License to your work, attach the following
630
- boilerplate notice, with the fields enclosed by brackets "[]"
631
- replaced with your own identifying information. (Don't include
632
- the brackets!) The text should be enclosed in the appropriate
633
- comment syntax for the file format. We also recommend that a
634
- file or class name and description of purpose be included on the
635
- same "printed page" as the copyright notice for easier
636
- identification within third-party archives.
637
-
638
- Copyright [yyyy] [name of copyright owner]
639
-
640
- Licensed under the Apache License, Version 2.0 (the "License");
641
- you may not use this file except in compliance with the License.
642
- You may obtain a copy of the License at
643
-
644
- http://www.apache.org/licenses/LICENSE-2.0
645
-
646
- Unless required by applicable law or agreed to in writing, software
647
- distributed under the License is distributed on an "AS IS" BASIS,
648
- WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
649
- See the License for the specific language governing permissions and
650
- limitations under the License.
651
- ```
652
-
653
- ### Citation Information
654
-
655
- Please cite the following paper when using this dataset.
656
-
657
- ```latex
658
- @misc{fitzgerald2022massive,
659
- title={MASSIVE: A 1M-Example Multilingual Natural Language Understanding Dataset with 51 Typologically-Diverse Languages},
660
- author={Jack FitzGerald and Christopher Hench and Charith Peris and Scott Mackie and Kay Rottmann and Ana Sanchez and Aaron Nash and Liam Urbach and Vishesh Kakarala and Richa Singh and Swetha Ranganath and Laurie Crist and Misha Britan and Wouter Leeuwis and Gokhan Tur and Prem Natarajan},
661
- year={2022},
662
- eprint={2204.08582},
663
- archivePrefix={arXiv},
664
- primaryClass={cs.CL}
665
- }
666
-
667
- @inproceedings{bastianelli-etal-2020-slurp,
668
- title = "{SLURP}: A Spoken Language Understanding Resource Package",
669
- author = "Bastianelli, Emanuele and
670
- Vanzo, Andrea and
671
- Swietojanski, Pawel and
672
- Rieser, Verena",
673
- booktitle = "Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP)",
674
- month = nov,
675
- year = "2020",
676
- address = "Online",
677
- publisher = "Association for Computational Linguistics",
678
- url = "https://aclanthology.org/2020.emnlp-main.588",
679
- doi = "10.18653/v1/2020.emnlp-main.588",
680
- pages = "7252--7262",
681
- abstract = "Spoken Language Understanding infers semantic meaning directly from audio data, and thus promises to reduce error propagation and misunderstandings in end-user applications. However, publicly available SLU resources are limited. In this paper, we release SLURP, a new SLU package containing the following: (1) A new challenging dataset in English spanning 18 domains, which is substantially bigger and linguistically more diverse than existing datasets; (2) Competitive baselines based on state-of-the-art NLU and ASR systems; (3) A new transparent metric for entity labelling which enables a detailed error analysis for identifying potential areas of improvement. SLURP is available at https://github.com/pswietojanski/slurp."
682
- }
683
- ```
 
1
+ ---
2
+ annotations_creators:
3
+ - machine-generated
4
+ - expert-generated
5
+ language_creators:
6
+ - found
7
+ languages:
8
+ - af-ZA
9
+ - am-ET
10
+ - ar-SA
11
+ - az-AZ
12
+ - bn-BD
13
+ - cy-GB
14
+ - da-DK
15
+ - de-DE
16
+ - el-GR
17
+ - en-US
18
+ - es-ES
19
+ - fa-IR
20
+ - fi-FI
21
+ - fr-FR
22
+ - he-IL
23
+ - hi-IN
24
+ - hu-HU
25
+ - hy-AM
26
+ - id-ID
27
+ - is-IS
28
+ - it-IT
29
+ - ja-JP
30
+ - jv-ID
31
+ - ka-GE
32
+ - km-KH
33
+ - kn-IN
34
+ - ko-KR
35
+ - lv-LV
36
+ - ml-IN
37
+ - mn-MN
38
+ - ms-MY
39
+ - my-MM
40
+ - nb-NO
41
+ - nl-NL
42
+ - pl-PL
43
+ - pt-PT
44
+ - ro-RO
45
+ - ru-RU
46
+ - sl-SL
47
+ - sq-AL
48
+ - sv-SE
49
+ - sw-KE
50
+ - ta-IN
51
+ - te-IN
52
+ - th-TH
53
+ - tl-PH
54
+ - tr-TR
55
+ - ur-PK
56
+ - vi-VN
57
+ - zh-CN
58
+ - zh-TW
59
+ licenses:
60
+ - apache-2.0
61
+ multilinguality:
62
+ - af-ZA
63
+ - am-ET
64
+ - ar-SA
65
+ - az-AZ
66
+ - bn-BD
67
+ - cy-GB
68
+ - da-DK
69
+ - de-DE
70
+ - el-GR
71
+ - en-US
72
+ - es-ES
73
+ - fa-IR
74
+ - fi-FI
75
+ - fr-FR
76
+ - he-IL
77
+ - hi-IN
78
+ - hu-HU
79
+ - hy-AM
80
+ - id-ID
81
+ - is-IS
82
+ - it-IT
83
+ - ja-JP
84
+ - jv-ID
85
+ - ka-GE
86
+ - km-KH
87
+ - kn-IN
88
+ - ko-KR
89
+ - lv-LV
90
+ - ml-IN
91
+ - mn-MN
92
+ - ms-MY
93
+ - my-MM
94
+ - nb-NO
95
+ - nl-NL
96
+ - pl-PL
97
+ - pt-PT
98
+ - ro-RO
99
+ - ru-RU
100
+ - sl-SL
101
+ - sq-AL
102
+ - sv-SE
103
+ - sw-KE
104
+ - ta-IN
105
+ - te-IN
106
+ - th-TH
107
+ - tl-PH
108
+ - tr-TR
109
+ - ur-PK
110
+ - vi-VN
111
+ - zh-CN
112
+ - zh-TW
113
+ pretty_name: MASSIVE
114
+ size_categories:
115
+ - 100K<n<1M
116
+ source_datasets:
117
+ - original
118
+ task_categories:
119
+ - text-classification
120
+ task_ids:
121
+ - intent-classification
122
+ - multi-class-classification
123
+ - natural-language-understanding
124
+ ---
125
+
126
+ # MASSIVE: A 1M-Example Multilingual Natural Language Understanding Dataset with 51 Typologically-Diverse Languages
127
+
128
+ ## Table of Contents
129
+ - [Dataset Card for [Needs More Information]](#dataset-card-for-needs-more-information)
130
+ - [Table of Contents](#table-of-contents)
131
+ - [Dataset Description](#dataset-description)
132
+ - [Dataset Summary](#dataset-summary)
133
+ - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards)
134
+ - [Languages](#languages)
135
+ - [Dataset Structure](#dataset-structure)
136
+ - [Data Instances](#data-instances)
137
+ - [Data Fields](#data-fields)
138
+ - [Data Splits](#data-splits)
139
+ - [Dataset Creation](#dataset-creation)
140
+ - [Curation Rationale](#curation-rationale)
141
+ - [Source Data](#source-data)
142
+ - [Initial Data Collection and Normalization](#initial-data-collection-and-normalization)
143
+ - [Who are the source language producers?](#who-are-the-source-language-producers)
144
+ - [Personal and Sensitive Information](#personal-and-sensitive-information)
145
+ - [Considerations for Using the Data](#considerations-for-using-the-data)
146
+ - [Other Known Limitations](#other-known-limitations)
147
+ - [Additional Information](#additional-information)
148
+ - [Dataset Curators](#dataset-curators)
149
+ - [Licensing Information](#licensing-information)
150
+ - [No Warranty](#no-warranty)
151
+ - [Citation Information](#citation-information)
152
+
153
+ ## Dataset Description
154
+
155
+ - **Homepage:** https://github.com/alexa/massive
156
+ - **Repository:** https://github.com/alexa/massive
157
+ - **Paper:** https://arxiv.org/abs/2204.08582
158
+ - **Leaderboard:** https://eval.ai/web/challenges/challenge-page/1697/overview
159
+ - **Point of Contact:** [GitHub](https://github.com/alexa/massive/issues)
160
+
161
+ ### Dataset Summary
162
+
163
+ MASSIVE is a parallel dataset of > 1M utterances across 51 languages with annotations for the Natural Language Understanding tasks of intent prediction and slot annotation. Utterances span 60 intents and include 55 slot types. MASSIVE was created by localizing the SLURP dataset, composed of general Intelligent Voice Assistant single-shot interactions.
164
+
165
+ | Name | Lang | Utt/Lang | Domains | Intents | Slots |
166
+ |:-------------------------------------------------------------------------------:|:-------:|:--------------:|:-------:|:--------:|:------:|
167
+ | MASSIVE | 51 | 19,521 | 18 | 60 | 55 |
168
+ | SLURP (Bastianelli et al., 2020) | 1 | 16,521 | 18 | 60 | 55 |
169
+ | NLU Evaluation Data (Liu et al., 2019) | 1 | 25,716 | 18 | 54 | 56 |
170
+ | Airline Travel Information System (ATIS) (Price, 1990) | 1 | 5,871 | 1 | 26 | 129 |
171
+ | ATIS with Hindi and Turkish (Upadhyay et al., 2018) | 3 | 1,315-5,871 | 1 | 26 | 129 |
172
+ | MultiATIS++ (Xu et al., 2020) | 9 | 1,422-5,897 | 1 | 21-26 | 99-140 |
173
+ | Snips (Coucke et al., 2018) | 1 | 14,484 | - | 7 | 53 |
174
+ | Snips with French (Saade et al., 2019) | 2 | 4,818 | 2 | 14-15 | 11-12 |
175
+ | Task Oriented Parsing (TOP) (Gupta et al., 2018) | 1 | 44,873 | 2 | 25 | 36 |
176
+ | Multilingual Task-Oriented Semantic Parsing (MTOP) (Li et al., 2021) | 6 | 15,195-22,288 | 11 | 104-113 | 72-75 |
177
+ | Cross-Lingual Multilingual Task Oriented Dialog (Schuster et al., 2019) | 3 | 5,083-43,323 | 3 | 12 | 11 |
178
+ | Microsoft Dialog Challenge (Li et al., 2018) | 1 | 38,276 | 3 | 11 | 29 |
179
+ | Fluent Speech Commands (FSC) (Lugosch et al., 2019) | 1 | 30,043 | - | 31 | - |
180
+ | Chinese Audio-Textual Spoken Language Understanding (CATSLU) (Zhu et al., 2019) | 1 | 16,258 | 4 | - | 94 |
181
+
182
+ ### Supported Tasks and Leaderboards
183
+
184
+ The dataset can be used to train a model for `natural-language-understanding` (NLU) :
185
+
186
+ - `intent-classification`
187
+ - `multi-class-classification`
188
+ - `natural-language-understanding`
189
+
190
+ ### Languages
191
+
192
+ The corpora consists of parallel sentences from 51 languages :
193
+
194
+ - `Afrikaans - South Africa (af-ZA)`
195
+ - `Amharic - Ethiopia (am-ET)`
196
+ - `Arabic - Saudi Arabia (ar-SA)`
197
+ - `Azeri - Azerbaijan (az-AZ)`
198
+ - `Bengali - Bangladesh (bn-BD)`
199
+ - `Chinese - China (zh-CN)`
200
+ - `Chinese - Taiwan (zh-TW)`
201
+ - `Danish - Denmark (da-DK)`
202
+ - `German - Germany (de-DE)`
203
+ - `Greek - Greece (el-GR)`
204
+ - `English - United States (en-US)`
205
+ - `Spanish - Spain (es-ES)`
206
+ - `Farsi - Iran (fa-IR)`
207
+ - `Finnish - Finland (fi-FI)`
208
+ - `French - France (fr-FR)`
209
+ - `Hebrew - Israel (he-IL)`
210
+ - `Hungarian - Hungary (hu-HU)`
211
+ - `Armenian - Armenia (hy-AM)`
212
+ - `Indonesian - Indonesia (id-ID)`
213
+ - `Icelandic - Iceland (is-IS)`
214
+ - `Italian - Italy (it-IT)`
215
+ - `Japanese - Japan (ja-JP)`
216
+ - `Javanese - Indonesia (jv-ID)`
217
+ - `Georgian - Georgia (ka-GE)`
218
+ - `Khmer - Cambodia (km-KH)`
219
+ - `Korean - Korea (ko-KR)`
220
+ - `Latvian - Latvia (lv-LV)`
221
+ - `Mongolian - Mongolia (mn-MN)`
222
+ - `Malay - Malaysia (ms-MY)`
223
+ - `Burmese - Myanmar (my-MM)`
224
+ - `Norwegian - Norway (nb-NO)`
225
+ - `Dutch - Netherlands (nl-NL)`
226
+ - `Polish - Poland (pl-PL)`
227
+ - `Portuguese - Portugal (pt-PT)`
228
+ - `Romanian - Romania (ro-RO)`
229
+ - `Russian - Russia (ru-RU)`
230
+ - `Slovanian - Slovania (sl-SL)`
231
+ - `Albanian - Albania (sq-AL)`
232
+ - `Swedish - Sweden (sv-SE)`
233
+ - `Swahili - Kenya (sw-KE)`
234
+ - `Hindi - India (hi-IN)`
235
+ - `Kannada - India (kn-IN)`
236
+ - `Malayalam - India (ml-IN)`
237
+ - `Tamil - India (ta-IN)`
238
+ - `Telugu - India (te-IN)`
239
+ - `Thai - Thailand (th-TH)`
240
+ - `Tagalog - Philippines (tl-PH)`
241
+ - `Turkish - Turkey (tr-TR)`
242
+ - `Urdu - Pakistan (ur-PK)`
243
+ - `Vietnamese - Vietnam (vi-VN)`
244
+ - `Welsh - United Kingdom (cy-GB)`
245
+
246
+ ## Load the dataset with HuggingFace
247
+
248
+ ```python
249
+ from datasets import load_dataset
250
+ dataset = load_dataset("qanastek/MASSIVE", "en-US", split='train')
251
+ print(dataset)
252
+ print(dataset[0])
253
+ ```
254
+
255
+ ## Dataset Structure
256
+
257
+ ### Data Instances (taken from Alexa Github)
258
+
259
+ ```json
260
+ {
261
+ "id": "0",
262
+ "locale": "de-DE",
263
+ "partition": "test",
264
+ "scenario": "alarm",
265
+ "intent": "alarm_set",
266
+ "utt": "weck mich diese woche um fünf uhr morgens auf",
267
+ "annot_utt": "weck mich [date : diese woche] um [time : fünf uhr morgens] auf",
268
+ "worker_id": "8",
269
+ "slot_method": [
270
+ {
271
+ "slot": "time",
272
+ "method": "translation"
273
+ },
274
+ {
275
+ "slot": "date",
276
+ "method": "translation"
277
+ }
278
+ ],
279
+ "judgments": [
280
+ {
281
+ "worker_id": "32",
282
+ "intent_score": 1,
283
+ "slots_score": 0,
284
+ "grammar_score": 4,
285
+ "spelling_score": 2,
286
+ "language_identification": "target"
287
+ },
288
+ {
289
+ "worker_id": "8",
290
+ "intent_score": 1,
291
+ "slots_score": 1,
292
+ "grammar_score": 4,
293
+ "spelling_score": 2,
294
+ "language_identification": "target"
295
+ },
296
+ {
297
+ "worker_id": "28",
298
+ "intent_score": 1,
299
+ "slots_score": 1,
300
+ "grammar_score": 4,
301
+ "spelling_score": 2,
302
+ "language_identification": "target"
303
+ }
304
+ ]
305
+ }
306
+ ```
307
+
308
+ ### Data Fields (taken from Alexa Github)
309
+
310
+ `id`: maps to the original ID in the [SLURP](https://github.com/pswietojanski/slurp) collection. Mapping back to the SLURP en-US utterance, this utterance served as the basis for this localization.
311
+
312
+ `locale`: is the language and country code accoring to ISO-639-1 and ISO-3166.
313
+
314
+ `partition`: is either `train`, `dev`, or `test`, according to the original split in [SLURP](https://github.com/pswietojanski/slurp).
315
+
316
+ `scenario`: is the general domain, aka "scenario" in SLURP terminology, of an utterance
317
+
318
+ `intent`: is the specific intent of an utterance within a domain formatted as `{scenario}_{intent}`
319
+
320
+ `utt`: the raw utterance text without annotations
321
+
322
+ `annot_utt`: the text from `utt` with slot annotations formatted as `[{label} : {entity}]`
323
+
324
+ `worker_id`: The obfuscated worker ID from MTurk of the worker completing the localization of the utterance. Worker IDs are specific to a locale and do *not* map across locales.
325
+
326
+ `slot_method`: for each slot in the utterance, whether that slot was a `translation` (i.e., same expression just in the target language), `localization` (i.e., not the same expression but a different expression was chosen more suitable to the phrase in that locale), or `unchanged` (i.e., the original en-US slot value was copied over without modification).
327
+
328
+ `judgments`: Each judgment collected for the localized utterance has 6 keys. `worker_id` is the obfuscated worker ID from MTurk of the worker completing the judgment. Worker IDs are specific to a locale and do *not* map across locales, but *are* consistent across the localization tasks and the judgment tasks, e.g., judgment worker ID 32 in the example above may appear as the localization worker ID for the localization of a different de-DE utterance, in which case it would be the same worker.
329
+
330
+ ```plain
331
+ intent_score : "Does the sentence match the intent?"
332
+ 0: No
333
+ 1: Yes
334
+ 2: It is a reasonable interpretation of the goal
335
+
336
+ slots_score : "Do all these terms match the categories in square brackets?"
337
+ 0: No
338
+ 1: Yes
339
+ 2: There are no words in square brackets (utterance without a slot)
340
+
341
+ grammar_score : "Read the sentence out loud. Ignore any spelling, punctuation, or capitalization errors. Does it sound natural?"
342
+ 0: Completely unnatural (nonsensical, cannot be understood at all)
343
+ 1: Severe errors (the meaning cannot be understood and doesn't sound natural in your language)
344
+ 2: Some errors (the meaning can be understood but it doesn't sound natural in your language)
345
+ 3: Good enough (easily understood and sounds almost natural in your language)
346
+ 4: Perfect (sounds natural in your language)
347
+
348
+ spelling_score : "Are all words spelled correctly? Ignore any spelling variances that may be due to differences in dialect. Missing spaces should be marked as a spelling error."
349
+ 0: There are more than 2 spelling errors
350
+ 1: There are 1-2 spelling errors
351
+ 2: All words are spelled correctly
352
+
353
+ language_identification : "The following sentence contains words in the following languages (check all that apply)"
354
+ 1: target
355
+ 2: english
356
+ 3: other
357
+ 4: target & english
358
+ 5: target & other
359
+ 6: english & other
360
+ 7: target & english & other
361
+ ```
362
+
363
+ ### Data Splits
364
+
365
+ |Language|Train|Dev|Test|
366
+ |:---:|:---:|:---:|:---:|
367
+ |af-ZA|11514|2033|2974|
368
+ |am-ET|11514|2033|2974|
369
+ |ar-SA|11514|2033|2974|
370
+ |az-AZ|11514|2033|2974|
371
+ |bn-BD|11514|2033|2974|
372
+ |cy-GB|11514|2033|2974|
373
+ |da-DK|11514|2033|2974|
374
+ |de-DE|11514|2033|2974|
375
+ |el-GR|11514|2033|2974|
376
+ |en-US|11514|2033|2974|
377
+ |es-ES|11514|2033|2974|
378
+ |fa-IR|11514|2033|2974|
379
+ |fi-FI|11514|2033|2974|
380
+ |fr-FR|11514|2033|2974|
381
+ |he-IL|11514|2033|2974|
382
+ |hi-IN|11514|2033|2974|
383
+ |hu-HU|11514|2033|2974|
384
+ |hy-AM|11514|2033|2974|
385
+ |id-ID|11514|2033|2974|
386
+ |is-IS|11514|2033|2974|
387
+ |it-IT|11514|2033|2974|
388
+ |ja-JP|11514|2033|2974|
389
+ |jv-ID|11514|2033|2974|
390
+ |ka-GE|11514|2033|2974|
391
+ |km-KH|11514|2033|2974|
392
+ |kn-IN|11514|2033|2974|
393
+ |ko-KR|11514|2033|2974|
394
+ |lv-LV|11514|2033|2974|
395
+ |ml-IN|11514|2033|2974|
396
+ |mn-MN|11514|2033|2974|
397
+ |ms-MY|11514|2033|2974|
398
+ |my-MM|11514|2033|2974|
399
+ |nb-NO|11514|2033|2974|
400
+ |nl-NL|11514|2033|2974|
401
+ |pl-PL|11514|2033|2974|
402
+ |pt-PT|11514|2033|2974|
403
+ |ro-RO|11514|2033|2974|
404
+ |ru-RU|11514|2033|2974|
405
+ |sl-SL|11514|2033|2974|
406
+ |sq-AL|11514|2033|2974|
407
+ |sv-SE|11514|2033|2974|
408
+ |sw-KE|11514|2033|2974|
409
+ |ta-IN|11514|2033|2974|
410
+ |te-IN|11514|2033|2974|
411
+ |th-TH|11514|2033|2974|
412
+ |tl-PH|11514|2033|2974|
413
+ |tr-TR|11514|2033|2974|
414
+ |ur-PK|11514|2033|2974|
415
+ |vi-VN|11514|2033|2974|
416
+ |zh-CN|11514|2033|2974|
417
+ |zh-TW|11514|2033|2974|
418
+
419
+ ## Dataset Creation
420
+
421
+ ### Source Data
422
+
423
+ #### Who are the source language producers?
424
+
425
+ The corpus has been produced and uploaded by Amazon Alexa.
426
+
427
+ ### Personal and Sensitive Information
428
+
429
+ The corpora is free of personal or sensitive information.
430
+
431
+ ## Additional Information
432
+
433
+ ### Dataset Curators
434
+
435
+ __MASSIVE__: Jack FitzGerald and Christopher Hench and Charith Peris and Scott Mackie and Kay Rottmann and Ana Sanchez and Aaron Nash and Liam Urbach and Vishesh Kakarala and Richa Singh and Swetha Ranganath and Laurie Crist and Misha Britan and Wouter Leeuwis and Gokhan Tur and Prem Natarajan.
436
+
437
+ __SLURP__: Bastianelli, Emanuele and Vanzo, Andrea and Swietojanski, Pawel and Rieser, Verena.
438
+
439
+ __Hugging Face__: Labrak Yanis (Not affiliated with the original corpus)
440
+
441
+ ### Licensing Information
442
+
443
+ ```plain
444
+ Copyright Amazon.com Inc. or its affiliates.
445
+
446
+ Copyright and license details for the data and modified code can be found in NOTICE.md.
447
+
448
+ License for massive repo and code, Apache 2.0:
449
+
450
+ Apache License
451
+ Version 2.0, January 2004
452
+ http://www.apache.org/licenses/
453
+
454
+ TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
455
+
456
+ 1. Definitions.
457
+
458
+ "License" shall mean the terms and conditions for use, reproduction,
459
+ and distribution as defined by Sections 1 through 9 of this document.
460
+
461
+ "Licensor" shall mean the copyright owner or entity authorized by
462
+ the copyright owner that is granting the License.
463
+
464
+ "Legal Entity" shall mean the union of the acting entity and all
465
+ other entities that control, are controlled by, or are under common
466
+ control with that entity. For the purposes of this definition,
467
+ "control" means (i) the power, direct or indirect, to cause the
468
+ direction or management of such entity, whether by contract or
469
+ otherwise, or (ii) ownership of fifty percent (50%) or more of the
470
+ outstanding shares, or (iii) beneficial ownership of such entity.
471
+
472
+ "You" (or "Your") shall mean an individual or Legal Entity
473
+ exercising permissions granted by this License.
474
+
475
+ "Source" form shall mean the preferred form for making modifications,
476
+ including but not limited to software source code, documentation
477
+ source, and configuration files.
478
+
479
+ "Object" form shall mean any form resulting from mechanical
480
+ transformation or translation of a Source form, including but
481
+ not limited to compiled object code, generated documentation,
482
+ and conversions to other media types.
483
+
484
+ "Work" shall mean the work of authorship, whether in Source or
485
+ Object form, made available under the License, as indicated by a
486
+ copyright notice that is included in or attached to the work
487
+ (an example is provided in the Appendix below).
488
+
489
+ "Derivative Works" shall mean any work, whether in Source or Object
490
+ form, that is based on (or derived from) the Work and for which the
491
+ editorial revisions, annotations, elaborations, or other modifications
492
+ represent, as a whole, an original work of authorship. For the purposes
493
+ of this License, Derivative Works shall not include works that remain
494
+ separable from, or merely link (or bind by name) to the interfaces of,
495
+ the Work and Derivative Works thereof.
496
+
497
+ "Contribution" shall mean any work of authorship, including
498
+ the original version of the Work and any modifications or additions
499
+ to that Work or Derivative Works thereof, that is intentionally
500
+ submitted to Licensor for inclusion in the Work by the copyright owner
501
+ or by an individual or Legal Entity authorized to submit on behalf of
502
+ the copyright owner. For the purposes of this definition, "submitted"
503
+ means any form of electronic, verbal, or written communication sent
504
+ to the Licensor or its representatives, including but not limited to
505
+ communication on electronic mailing lists, source code control systems,
506
+ and issue tracking systems that are managed by, or on behalf of, the
507
+ Licensor for the purpose of discussing and improving the Work, but
508
+ excluding communication that is conspicuously marked or otherwise
509
+ designated in writing by the copyright owner as "Not a Contribution."
510
+
511
+ "Contributor" shall mean Licensor and any individual or Legal Entity
512
+ on behalf of whom a Contribution has been received by Licensor and
513
+ subsequently incorporated within the Work.
514
+
515
+ 2. Grant of Copyright License. Subject to the terms and conditions of
516
+ this License, each Contributor hereby grants to You a perpetual,
517
+ worldwide, non-exclusive, no-charge, royalty-free, irrevocable
518
+ copyright license to reproduce, prepare Derivative Works of,
519
+ publicly display, publicly perform, sublicense, and distribute the
520
+ Work and such Derivative Works in Source or Object form.
521
+
522
+ 3. Grant of Patent License. Subject to the terms and conditions of
523
+ this License, each Contributor hereby grants to You a perpetual,
524
+ worldwide, non-exclusive, no-charge, royalty-free, irrevocable
525
+ (except as stated in this section) patent license to make, have made,
526
+ use, offer to sell, sell, import, and otherwise transfer the Work,
527
+ where such license applies only to those patent claims licensable
528
+ by such Contributor that are necessarily infringed by their
529
+ Contribution(s) alone or by combination of their Contribution(s)
530
+ with the Work to which such Contribution(s) was submitted. If You
531
+ institute patent litigation against any entity (including a
532
+ cross-claim or counterclaim in a lawsuit) alleging that the Work
533
+ or a Contribution incorporated within the Work constitutes direct
534
+ or contributory patent infringement, then any patent licenses
535
+ granted to You under this License for that Work shall terminate
536
+ as of the date such litigation is filed.
537
+
538
+ 4. Redistribution. You may reproduce and distribute copies of the
539
+ Work or Derivative Works thereof in any medium, with or without
540
+ modifications, and in Source or Object form, provided that You
541
+ meet the following conditions:
542
+
543
+ (a) You must give any other recipients of the Work or
544
+ Derivative Works a copy of this License; and
545
+
546
+ (b) You must cause any modified files to carry prominent notices
547
+ stating that You changed the files; and
548
+
549
+ (c) You must retain, in the Source form of any Derivative Works
550
+ that You distribute, all copyright, patent, trademark, and
551
+ attribution notices from the Source form of the Work,
552
+ excluding those notices that do not pertain to any part of
553
+ the Derivative Works; and
554
+
555
+ (d) If the Work includes a "NOTICE" text file as part of its
556
+ distribution, then any Derivative Works that You distribute must
557
+ include a readable copy of the attribution notices contained
558
+ within such NOTICE file, excluding those notices that do not
559
+ pertain to any part of the Derivative Works, in at least one
560
+ of the following places: within a NOTICE text file distributed
561
+ as part of the Derivative Works; within the Source form or
562
+ documentation, if provided along with the Derivative Works; or,
563
+ within a display generated by the Derivative Works, if and
564
+ wherever such third-party notices normally appear. The contents
565
+ of the NOTICE file are for informational purposes only and
566
+ do not modify the License. You may add Your own attribution
567
+ notices within Derivative Works that You distribute, alongside
568
+ or as an addendum to the NOTICE text from the Work, provided
569
+ that such additional attribution notices cannot be construed
570
+ as modifying the License.
571
+
572
+ You may add Your own copyright statement to Your modifications and
573
+ may provide additional or different license terms and conditions
574
+ for use, reproduction, or distribution of Your modifications, or
575
+ for any such Derivative Works as a whole, provided Your use,
576
+ reproduction, and distribution of the Work otherwise complies with
577
+ the conditions stated in this License.
578
+
579
+ 5. Submission of Contributions. Unless You explicitly state otherwise,
580
+ any Contribution intentionally submitted for inclusion in the Work
581
+ by You to the Licensor shall be under the terms and conditions of
582
+ this License, without any additional terms or conditions.
583
+ Notwithstanding the above, nothing herein shall supersede or modify
584
+ the terms of any separate license agreement you may have executed
585
+ with Licensor regarding such Contributions.
586
+
587
+ 6. Trademarks. This License does not grant permission to use the trade
588
+ names, trademarks, service marks, or product names of the Licensor,
589
+ except as required for reasonable and customary use in describing the
590
+ origin of the Work and reproducing the content of the NOTICE file.
591
+
592
+ 7. Disclaimer of Warranty. Unless required by applicable law or
593
+ agreed to in writing, Licensor provides the Work (and each
594
+ Contributor provides its Contributions) on an "AS IS" BASIS,
595
+ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
596
+ implied, including, without limitation, any warranties or conditions
597
+ of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
598
+ PARTICULAR PURPOSE. You are solely responsible for determining the
599
+ appropriateness of using or redistributing the Work and assume any
600
+ risks associated with Your exercise of permissions under this License.
601
+
602
+ 8. Limitation of Liability. In no event and under no legal theory,
603
+ whether in tort (including negligence), contract, or otherwise,
604
+ unless required by applicable law (such as deliberate and grossly
605
+ negligent acts) or agreed to in writing, shall any Contributor be
606
+ liable to You for damages, including any direct, indirect, special,
607
+ incidental, or consequential damages of any character arising as a
608
+ result of this License or out of the use or inability to use the
609
+ Work (including but not limited to damages for loss of goodwill,
610
+ work stoppage, computer failure or malfunction, or any and all
611
+ other commercial damages or losses), even if such Contributor
612
+ has been advised of the possibility of such damages.
613
+
614
+ 9. Accepting Warranty or Additional Liability. While redistributing
615
+ the Work or Derivative Works thereof, You may choose to offer,
616
+ and charge a fee for, acceptance of support, warranty, indemnity,
617
+ or other liability obligations and/or rights consistent with this
618
+ License. However, in accepting such obligations, You may act only
619
+ on Your own behalf and on Your sole responsibility, not on behalf
620
+ of any other Contributor, and only if You agree to indemnify,
621
+ defend, and hold each Contributor harmless for any liability
622
+ incurred by, or claims asserted against, such Contributor by reason
623
+ of your accepting any such warranty or additional liability.
624
+
625
+ END OF TERMS AND CONDITIONS
626
+
627
+ APPENDIX: How to apply the Apache License to your work.
628
+
629
+ To apply the Apache License to your work, attach the following
630
+ boilerplate notice, with the fields enclosed by brackets "[]"
631
+ replaced with your own identifying information. (Don't include
632
+ the brackets!) The text should be enclosed in the appropriate
633
+ comment syntax for the file format. We also recommend that a
634
+ file or class name and description of purpose be included on the
635
+ same "printed page" as the copyright notice for easier
636
+ identification within third-party archives.
637
+
638
+ Copyright [yyyy] [name of copyright owner]
639
+
640
+ Licensed under the Apache License, Version 2.0 (the "License");
641
+ you may not use this file except in compliance with the License.
642
+ You may obtain a copy of the License at
643
+
644
+ http://www.apache.org/licenses/LICENSE-2.0
645
+
646
+ Unless required by applicable law or agreed to in writing, software
647
+ distributed under the License is distributed on an "AS IS" BASIS,
648
+ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
649
+ See the License for the specific language governing permissions and
650
+ limitations under the License.
651
+ ```
652
+
653
+ ### Citation Information
654
+
655
+ Please cite the following paper when using this dataset.
656
+
657
+ ```latex
658
+ @misc{fitzgerald2022massive,
659
+ title={MASSIVE: A 1M-Example Multilingual Natural Language Understanding Dataset with 51 Typologically-Diverse Languages},
660
+ author={Jack FitzGerald and Christopher Hench and Charith Peris and Scott Mackie and Kay Rottmann and Ana Sanchez and Aaron Nash and Liam Urbach and Vishesh Kakarala and Richa Singh and Swetha Ranganath and Laurie Crist and Misha Britan and Wouter Leeuwis and Gokhan Tur and Prem Natarajan},
661
+ year={2022},
662
+ eprint={2204.08582},
663
+ archivePrefix={arXiv},
664
+ primaryClass={cs.CL}
665
+ }
666
+
667
+ @inproceedings{bastianelli-etal-2020-slurp,
668
+ title = "{SLURP}: A Spoken Language Understanding Resource Package",
669
+ author = "Bastianelli, Emanuele and
670
+ Vanzo, Andrea and
671
+ Swietojanski, Pawel and
672
+ Rieser, Verena",
673
+ booktitle = "Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP)",
674
+ month = nov,
675
+ year = "2020",
676
+ address = "Online",
677
+ publisher = "Association for Computational Linguistics",
678
+ url = "https://aclanthology.org/2020.emnlp-main.588",
679
+ doi = "10.18653/v1/2020.emnlp-main.588",
680
+ pages = "7252--7262",
681
+ abstract = "Spoken Language Understanding infers semantic meaning directly from audio data, and thus promises to reduce error propagation and misunderstandings in end-user applications. However, publicly available SLU resources are limited. In this paper, we release SLURP, a new SLU package containing the following: (1) A new challenging dataset in English spanning 18 domains, which is substantially bigger and linguistically more diverse than existing datasets; (2) Competitive baselines based on state-of-the-art NLU and ASR systems; (3) A new transparent metric for entity labelling which enables a detailed error analysis for identifying potential areas of improvement. SLURP is available at https://github.com/pswietojanski/slurp."
682
+ }
683
+ ```