Quality at a Glance: An Audit of Web-Crawled Multilingual Datasets
Julia Kreutzer
(1, 2)
,
Isaac Caswell
(1)
,
Lisa Wang
(1)
,
Ahsan Wahab
(3)
,
Daan van Esch
(1)
,
Nasanbayar Ulzii-Orshikh
(4)
,
Allahsera Tapo
(2, 5)
,
Nishant Subramani
(2, 6)
,
Artem Sokolov
(1)
,
Claytone Sikasote
(2, 7)
,
Monang Setyawan
(1)
,
Supheakmungkol Sarin
(3)
,
Sokhar Samb
(8)
,
Benoît Sagot
(9)
,
Clara Rivera
(1)
,
Annette Rios
(10)
,
Isabel Papadimitriou
(11)
,
Salomey Osei
(2, 12)
,
Pedro Ortiz Suarez
(9, 13)
,
Iroro Orife
(2)
,
Kelechi Ogueji
(2, 14)
,
Rubungo Andre Niyongabo
(2, 15)
,
Toan Q. Nguyen
(16)
,
Mathias Müller
(10)
,
André Müller
(10)
,
Shamsuddeen Hassan Muhammad
(2, 17)
,
Nanda Muhammad
(1)
,
Ayanda Mnyakeni
(1)
,
Jamshidbek Mirzakhalov
(3, 18)
,
Tapiwanashe Matangira
(1)
,
Colin Leong
(2)
,
Nze Lawson
(1)
,
Sneha Kudugunta
(1)
,
Yacine Jernite
(2, 19)
,
Mathias Jenny
(10)
,
Orhan Firat
(3, 1)
,
Bonaventure F. P. Dossou
(2, 20)
,
Sakhile Dlamini
(1)
,
Nisansa de Silva
(21)
,
Sakine Çabuk Balli
(1)
,
Stella Biderman
(22)
,
Alessia Battisti
(10)
,
Ahmed Baruwa
(2, 23)
,
Ankur Bapna
(1)
,
Pallavi Baljekar
(1)
,
Israel Abebe Azime
(2, 8)
,
Ayodele Awokoya
(2, 24)
,
Duygu Ataman
(3, 10)
,
Orevaoghene Ahia
(2, 25)
,
Oghenefego Ahia
(3)
,
Sweta Agrawal
(26)
,
Mofetoluwa Adeyemi
(2, 27)
1
Google Inc.
2 Masakhane NLP
3 TIL - Turkic Interlingua
4 Computer Science Department [Haveford]
5 RobotsMali
6 Intel Labs Berkeley
7 UNZA - University of Zambia [Lusaka]
8 AIMS - African Institute for Mathematical Sciences
9 ALMAnaCH - Automatic Language Modelling and ANAlysis & Computational Humanities
10 UZH - Universität Zürich [Zürich] = University of Zurich
11 Stanford University
12 KNUST - Kwame Nkrumah University of Science and Technology
13 SU - Sorbonne Université
14 University of Waterloo [Waterloo]
15 UESTC - University of Electronic Science and Technology of China [Chengdu]
16 UND - University of Notre Dame [Indiana]
17 BUK - Bayero University Kano
18 USF - University of South Florida [Tampa]
19 Hugging Face
20 Jacobs University = Constructor University [Bremen]
21 University of Moratuwa
22 EleutherAI
23 OAU - Obafemi Awolowo University
24 University of Ibadan
25 InstaDeep
26 University of Maryland [Baltimore]
27 Defence Space Administration [Abuja]
2 Masakhane NLP
3 TIL - Turkic Interlingua
4 Computer Science Department [Haveford]
5 RobotsMali
6 Intel Labs Berkeley
7 UNZA - University of Zambia [Lusaka]
8 AIMS - African Institute for Mathematical Sciences
9 ALMAnaCH - Automatic Language Modelling and ANAlysis & Computational Humanities
10 UZH - Universität Zürich [Zürich] = University of Zurich
11 Stanford University
12 KNUST - Kwame Nkrumah University of Science and Technology
13 SU - Sorbonne Université
14 University of Waterloo [Waterloo]
15 UESTC - University of Electronic Science and Technology of China [Chengdu]
16 UND - University of Notre Dame [Indiana]
17 BUK - Bayero University Kano
18 USF - University of South Florida [Tampa]
19 Hugging Face
20 Jacobs University = Constructor University [Bremen]
21 University of Moratuwa
22 EleutherAI
23 OAU - Obafemi Awolowo University
24 University of Ibadan
25 InstaDeep
26 University of Maryland [Baltimore]
27 Defence Space Administration [Abuja]
Benoît Sagot
- Fonction : Auteur
- PersonId : 1461
- IdHAL : bsagot
- ORCID : 0000-0002-0107-8526
- IdRef : 177454229
Pedro Ortiz Suarez
- Fonction : Auteur
- PersonId : 178412
- IdHAL : pedro-ortiz-suarez
- ORCID : 0000-0003-0343-8852
- IdRef : 264210743
Duygu Ataman
- Fonction : Auteur
Résumé
With the success of large-scale pre-training and multilingual modeling in Natural Language Processing (NLP), recent years have seen a proliferation of large, web-mined text datasets covering hundreds of languages. However, to date there has been no systematic analysis of the quality of these publicly available datasets, or whether the datasets actually contain content in the languages they claim to represent. In this work, we manually audit the quality of 205 language-specific corpora released with five major public datasets (CCAligned, ParaCrawl, WikiMatrix, OSCAR, mC4), and audit the correctness of language codes in a sixth (JW300). We find that lower-resource corpora have systematic issues: at least 15 corpora are completely erroneous, and a significant fraction contains less than 50% sentences of acceptable quality. Similarly, we find 82 corpora that are mislabeled or use nonstandard/ambiguous language codes. We demonstrate that these issues are easy to detect even for non-speakers of the languages in question, and supplement the human judgements with automatic analyses. Inspired by our analysis, we recommend techniques to evaluate and improve multilingual corpora and discuss the risks that come with low-quality data releases.
Domaines
Informatique et langage [cs.CL]| Origine | Fichiers produits par l'(les) auteur(s) |
|---|
