wikidata_id
stringlengths
3
10
lastrevid
int64
306M
1.56B
label
stringlengths
1
120
Q110425075
1,556,708,450
Lopata
Q109946856
1,556,662,944
Leschallier
Q110409516
1,556,627,503
Johann-Heinrich
Q108868778
1,556,622,179
Korndörfer
Q109411983
1,556,588,123
Padró
Q101214825
1,556,587,003
Akofa
Q106335056
1,556,559,942
Zemmel
Q110419747
1,556,542,153
Rickwart
Q110419795
1,556,536,253
Lapeyra
Q109942293
1,556,534,964
Preisigke
Q106113565
1,556,529,368
Rixová
Q96722160
1,556,526,608
Ri
Q110417577
1,556,511,658
Rüttinger
Q110419264
1,556,511,403
Haldimand
Q110419020
1,556,511,391
Bolau
Q110418692
1,556,511,340
Brodyaga
Q110418324
1,556,511,259
Máriássy
Q110418136
1,556,511,230
Fonssagrives
Q110418126
1,556,511,213
van de Vijver
Q110418025
1,556,511,112
Paufler
Q110417998
1,556,511,102
Zuijderwijk
Q110417588
1,556,511,091
Haguenin
Q110417455
1,556,509,939
Budinszky
Q110416022
1,556,509,922
Pongolini
Q110415825
1,556,509,906
Schipa
Q110415699
1,556,509,890
Imbruglia
Q110414976
1,556,509,872
Lo Muzio
Q110414968
1,556,509,787
Rokx
Q110414876
1,556,509,775
Socrate
Q110414777
1,556,509,765
Tonies
Q110414767
1,556,509,746
Geurts van Kessel
Q110414469
1,556,509,646
Speksnijder
Q110414412
1,556,509,611
van Lokhorst
Q110414387
1,556,509,506
Schryff
Q110414369
1,556,509,488
Roobol
Q110414347
1,556,509,479
Witvliet
Q110414321
1,556,509,463
Ter Laan
Q110414288
1,556,509,342
Van Den Ende
Q110360226
1,556,509,326
van den Ende
Q110414277
1,556,509,249
van der Ende
Q110414256
1,556,508,977
Van den Ende
Q110414006
1,556,508,740
Filomusi-Guelfi
Q110413453
1,556,508,726
Virieux
Q110412457
1,556,508,711
Mazanowski
Q110411993
1,556,508,606
Alzari
Q110411981
1,556,508,500
von Schröderß
Q110411789
1,556,507,410
Rockx
Q110411555
1,556,507,087
van Nierop
Q110410957
1,556,507,047
Allwork
Q110409748
1,556,507,025
Sinistrero
Q110409560
1,556,506,920
Gropelli
Q110409425
1,556,506,839
Kalniņa
Q110409304
1,556,506,784
Marylski
Q110408983
1,556,506,666
Bērziņa
Q110408656
1,556,506,582
Vialardi
Q110408258
1,556,506,464
Olorón
Q110408099
1,556,506,250
Ollo
Q110412203
1,556,504,098
Stoevesandt
Q110419005
1,556,503,968
Stövesand
Q101043160
1,556,503,837
Stöwsand
Q100587794
1,556,501,368
Stuyvesant
Q95852982
1,556,495,915
Christian Stövesand
Q110418823
1,556,484,715
von Opel
Q105955911
1,556,481,907
Marie
Q106307292
1,556,459,013
Ringvatnet
Q83338292
1,556,450,484
Abelen
Q83338286
1,556,450,362
Abbel
Q83338298
1,556,444,403
Abelius
Q109940969
1,556,436,850
Mvondo
Q109940779
1,556,436,737
Mbaka
Q97097207
1,556,399,212
Telegdi
Q89974784
1,556,368,257
Scherphuis
Q110418284
1,556,363,767
Obbo
Q104002197
1,556,344,354
Van van de Vijver
Q83374934
1,556,336,341
Oppler
Q83374933
1,556,327,443
Oppl
Q96600611
1,556,317,560
Gerhard von Glahn
Q83374622
1,556,301,101
Obelt
Q109937640
1,556,298,323
Ponjoan
Q110393330
1,556,293,609
Rips
Q110413372
1,556,237,147
Opelt-Stoevesandt
Q107423630
1,556,201,633
Robert Pfleger
Q96582228
1,556,181,126
Flahault
Q107203945
1,556,176,775
Abdulfatai Yahaya Seriki
Q109935462
1,556,173,700
Creuzet
Q79392884
1,556,165,966
Georgievna
Q109935312
1,556,164,751
İsmayılov
Q109934255
1,556,132,312
Poincignon
Q110381224
1,556,130,039
ter Laan
Q110414079
1,556,118,604
Costenco
Q110414104
1,556,118,414
Kostenko
Q110413768
1,556,116,777
Odbald
Q110386877
1,556,095,476
Wölbling
Q110334596
1,556,093,883
Huldermann
Q106832845
1,556,076,975
Hans Joachim Stoevesandt
Q109932031
1,556,049,005
Ferrerons
Q104621420
1,556,021,706
Sigismonda
Q63217071
1,555,994,396
Johan Erik
Q63689481
1,555,993,901
Joan Enric
Q110409460
1,555,989,548
Kenwrick

Wikidata Labels

Large parallel corpus for machine translation

  • Entity label data extracted from Wikidata (2022-01-03), filtered for item entities only
  • Only download the languages you need with datasets>=2.14.0
  • Similar dataset: https://huggingface.co/datasets/wmt/wikititles (18 Wikipedia titles pairs instead of all Wikidata entities)

Dataset Details

Dataset Sources

Uses

You can generate parallel text examples from this dataset like below:

from datasets import load_dataset
import pandas as pd

def parallel_labels(lang_codes: list, how="inner", repo_id="rayliuca/wikidata_entity_label", merge_config={}, datasets_config={}) -> pd.DataFrame:
    out_df = None
    for lc in lang_codes:
        dataset = load_dataset(repo_id, lc, **datasets_config)
        dataset_df = dataset['label'].to_pandas().rename(columns={"label":lc}).drop(columns=['lastrevid'])
        if out_df is None:
            out_df = dataset_df
        else:
            out_df = out_df.merge(
                    dataset_df,
                    on='wikidata_id',
                    how=how,
                    **merge_config
                )
    return out_df

# Note: the "en" subset is >4GB
parallel_labels(['en', 'fr', 'ja', 'zh']).head()

Output

wikidata_id en fr ja zh
0 Q109739412 SARS-CoV-2 Omicron variant variant Omicron du SARS-CoV-2 SARSコロナウイルス2-オミクロン株 嚴重急性呼吸道症候群冠狀病毒2型Omicron變異株
1 Q108460606 Ulughbegsaurus Ulughbegsaurus ウルグベグサウルス 兀魯伯龍屬
2 Q108556886 AUKUS AUKUS AUKUS AUKUS
3 Q106496152 Claude Joseph Claude Joseph クロード・ジョゼフ 克洛德·约瑟夫
4 Q105519361 The World's Finest Assassin Gets Reincarnated in a Different World as an Aristocrat The World's Finest Assassin Gets Reincarnated in Another World as an Aristocrat 世界最高の暗殺者、異世界貴族に転生する 世界頂尖的暗殺者轉生為異世界貴族

Note: this example table above shows a quirk(?) of the Wiki data. The French Wikipedia page The World's Finest Assassin Gets Reincarnated in Another World as an Aristocrat uses English for its title. While this could be disadvantageous for direct translation training, it also provides insights into how native speakers might call this entity instead of the literal translation on the Wiki page as well

Dataset Structure

Each language has its own subset (aka config), which means you only have to download the languages you need with datasets>=2.14.0

Each subset has these fields:

  • wikidata_id
  • lastrevid
  • label

Dataset Creation

Data Collection and Processing

  • Filtered for item entities only
  • Ignored the descriptions as those texts are not very parallel

Bias, Risks, and Limitations

  • Might be slightly outdated (2022)
  • Popular languages have more entries
  • Labels are not guaranteed to be literal translations (see examples above)
Downloads last month
10,211