MasakhaNER 2.0 Africa-centric Transfer Learning for Named Entity Recognition David Ifeoluwa Adelani12 Graham Neubig3 Sebastian Ruder4 Shruti Rijhwani3

2025-05-02 0 0 1.07MB 22 页 10玖币

侵权投诉

MasakhaNER 2.0: Africa-centric Transfer Learning for Named Entity

Recognition

David Ifeoluwa Adelani1,2,∗, Graham Neubig3, Sebastian Ruder4, Shruti Rijhwani3,

Michael Beukman5∗, Chester Palen-Michel6∗, Constantine Lignos6∗, Jesujoba O. Alabi1∗,

Shamsuddeen H. Muhammad7∗, Peter Nabende8∗, Cheikh M. Bamba Dione9∗, Andiswa Bukula10,

Rooweither Mabuya10 , Bonaventure F. P. Dossou11∗, Blessing Sibanda∗, Happy Buzaaba12∗,

Jonathan Mukiibi8∗, Godson Kalipe∗, Derguene Mbaye13∗, Amelia Taylor14∗, Fatoumata Kabore15∗,

Chris Chinenye Emezue16∗, Anuoluwapo Aremu∗, Perez Ogayo3∗, Catherine Gitau∗,

Edwin Munkoh-Buabeng17∗, Victoire M. Koagne∗, Allahsera Auguste Tapo18∗, Tebogo Macucwa19∗,

Vukosi Marivate19∗, Elvis Mboning∗, Tajuddeen Gwadabe∗, Tosin Adewumi20∗,

Orevaoghene Ahia21∗, Joyce Nakatumba-Nabende8∗, Neo L. Mokono19∗, Ignatius Ezeani22∗,

Chiamaka Chukwuneke22∗, Mofetoluwa Adeyemi23∗, Gilles Q. Hacheme24∗, Idris Abdulmumin25∗,

Odunayo Ogundepo23∗, Oreen Yousuf15∗, Tatiana Moteu Ngoli∗, Dietrich Klakow1

∗Masakhane NLP, 1Saarland University, Germany, 2University College London, UK, 3Carnegie Mellon University, USA,

4Google Research, 5University of the Witwatersrand, South Africa, 6Brandeis University, USA, 7LIAAD-INESC TEC, Portugal,

8Makerere University, Uganda 9University of Bergen, Norway, 10SADiLaR, South Africa, 11 Mila Quebec AI Institute, Canada,

12RIKEN Center for AI Project, Japan, 13 Baamtu, Senegal, 14Malawi University of Business and Applied Science, Malawi,

15Uppsala University, Sweden, 16TU Munich, Germany, 17TU Clausthal, Germany, 18Rochester Institute of Technology, USA,

19University of Pretoria, South Africa, 20Luleå University of Technology, Sweden, 21University of Washington, USA,

22Lancaster University, UK, 23University of Waterloo, Canada, 24Ai4innov, France, 25Ahmadu Bello University, Nigeria.

Abstract

African languages are spoken by over a bil-

lion people, but are underrepresented in NLP

research and development. The challenges im-

peding progress include the limited availabil-

ity of annotated datasets, as well as a lack

of understanding of the settings where cur-

rent methods are effective. In this paper, we

make progress towards solutions for these chal-

lenges, focusing on the task of named en-

tity recognition (NER). We create the largest

human-annotated NER dataset for 20 African

languages, and we study the behavior of state-

of-the-art cross-lingual transfer methods in an

Africa-centric setting, demonstrating that the

choice of source language signiﬁcantly affects

performance. We show that choosing the

best transfer language improves zero-shot F1

scores by an average of 14 points across 20

languages compared to using English. Our re-

sults highlight the need for benchmark datasets

and models that cover typologically-diverse

African languages.

1 Introduction

Many African languages are spoken by millions

or tens of millions of speakers. However, these

languages are poorly represented in NLP research,

and the development of NLP systems for African

languages is often limited by the lack of datasets

for training and evaluation (Adelani et al.,2021b).

Additionally, while there has been much re-

cent work in using zero-shot cross-lingual trans-

fer (Ponti et al.,2020;Pfeiffer et al.,2020;

Ebrahimi et al.,2022) to improve performance

on tasks for low-resource languages with multilin-

gual pretrained language models (PLMs) (Devlin

et al.,2019a;Conneau et al.,2020), the settings un-

der which contemporary transfer learning methods

work best are still unclear (Pruksachatkun et al.,

2020;Lauscher et al.,2020;Xia et al.,2020). For

example, several methods use English as the source

language because of the availability of training data

across many tasks (Hu et al.,2020;Ruder et al.,

2021), but there is evidence that English is often

not the best transfer language (Lin et al.,2019;

de Vries et al.,2022;Oladipo et al.,2022), and the

process of choosing the best source language to

transfer from remains an open question.

There has been recent progress in creating bench-

mark datasets for training and evaluating models

in African languages for several tasks such as ma-

chine translation (

∀

et al.,2020;Reid et al.,2021;

Adelani et al.,2021a,2022;Abdulmumin et al.,

2022), and sentiment analysis (Yimam et al.,2020;

Muhammad et al.,2022). In this paper, we focus on

the standard NLP task of named entity recognition

(NER) because of its utility in downstream applica-

tions such as question answering and information

arXiv:2210.12391v2 [cs.CL] 15 Nov 2022

extraction. For NER, annotated datasets exist only

in a few African languages (Adelani et al.,2021b;

Yohannes and Amagasa,2022), the largest of which

is the MasakhaNER dataset (Adelani et al.,2021b)

(which we call MasakhaNER 1.0 in the remainder

of the paper). While MasakhaNER 1.0 covers 10

African languages spoken mostly in West and East

Africa, it does not include any languages spoken

in Southern Africa, which have distinct syntactic

and morphological characteristics and are spoken

by 40 million people.

In this paper, we tackle two current challenges

in developing NER models for African languages:

(1) the lack of typologically- and geographically-

diverse evaluation datasets for African languages;

and (2) choosing the best transfer language for

NER in an Africa-centric setting, which has not

been previously explored in the literature.

To address the ﬁrst challenge, we create the

MasakhaNER 2.0 corpus, the largest human-

annotated NER dataset for African languages.

MasakhaNER 2.0 contains annotated text data from

20 languages widely spoken in Sub-Saharan Africa

and is complementary to the languages present in

previously existing datasets (e.g., Adelani et al.,

2021b). We discuss our annotation methodology

as well as perform benchmarking experiments on

our dataset with state-of-the-art NER models based

on multilingual PLMs.

In addition, to better understand the effect of

source language on transfer learning, we exten-

sively analyze different features that contribute to

cross-lingual transfer, including linguistic charac-

teristics of the languages (i.e., typological, geo-

graphical, and phylogenetic features) as well as

data-dependent features such as entity overlap

across source and target languages (Lin et al.,

2019). We demonstrate that choosing the best

transfer language(s) in both single-source and co-

training setups leads to large improvements in NER

performance in zero-shot settings; our experiments

show an average of a 14 point increase in F1 score

as compared to using English as source language

across 20 target African languages. We release the

data, code, and models on Github1

2 Related Work

African NER Datasets

There are some human-

annotated NER datasets for African languages

1https://github.com/masakhane-io/

masakhane-ner/tree/main/MasakhaNER2.0

such as the SaDiLAR NER corpus (Eiselen,2016)

covering 10 South African languages, LORELEI

(Strassel and Tracey,2016), which covers nine

African languages but is not open-sourced, and

some individual language efforts for Amharic (Jib-

ril and Tantug,2022), Yorùbá (Alabi et al.,

2020), Hausa (Hedderich et al.,2020), and

Tigrinya (Yohannes and Amagasa,2022). Closest

to our work is the MasakhaNER 1.0 corpus (Ade-

lani et al.,2021b), which covers 10 widely spo-

ken languages in the news domain, but excludes

languages from the southern region of Africa like

isiZulu, isiXhosa, and chiShona with distinct syn-

tactic features (e.g., noun preﬁxes and capitaliza-

tion in between words) which limits transfer learn-

ing from other languages. We include ﬁve lan-

guages from Southern Africa in our new corpus.

Cross-lingual Transfer

Leveraging cross-

lingual transfer has the potential to drastically

improve model performance without requiring

large amounts of data in the target language (Con-

neau et al.,2020) but it is not always clear from

which language we must transfer from (Lin et al.,

2019;de Vries et al.,2022). To this end, recent

work investigates methods for selecting good

transfer languages and informative features. For

instance, token overlap between the source and

target language is a useful predictor of transfer

performance for some tasks (Lin et al.,2019;

Wu and Dredze,2019). Linguistic distance (Lin

et al.,2019;de Vries et al.,2022), word order (K

et al.,2020;Pires et al.,2019) and script dif-

ferences (de Vries et al.,2022), and syntactic

similarity (Karamolegkou and Stymne,2021) have

also been shown to impact performance. Another

research direction attempts to build models of

transfer performance that predicts the best transfer

language for a target language by using some

linguistic and data-dependent features (Lin et al.,

2019;Ahuja et al.,2022).

3 Languages and Their Characteristics

3.1 Focus Languages

Table 1 provides an overview of the languages in

our MasakhaNER 2.0 corpus. We focus on 20 Sub-

Saharan African languages

with varying numbers

of speakers (between 1M–100M) that are spoken

by over 500M people in around 27 countries in

Our selection was also constrained by the availability

of volunteers that speak the languages in different NLP/AI

communities in Africa.

African No. of % Entities #

Language Family Region Speakers Source Train / dev / test in Tokens Tokens

Bambara (bam) NC / Mande West 14M MAFAND-MT (Adelani et al.,2022) 4462/ 638/ 1274 6.5 155,552

Ghomálá’ (bbj) NC / Grassﬁelds Central 1M MAFAND-MT (Adelani et al.,2022) 3384/ 483/ 966 11.3 69,474

Éwé (ewe) NC / Kwa West 7M MAFAND-MT (Adelani et al.,2022) 3505/ 501/ 1001 15.3 90420

Fon (fon) NC / Volta-Niger West 2M MAFAND-MT (Adelani et al.,2022) 4343/ 621/ 1240 8.3 173,099

Hausa (hau) Afro-Asiatic / Chadic West 63M Kano Focus and Freedom Radio 5716/ 816/ 1633 14.0 221,086

Igbo (ibo) NC / Volta-Niger West 27M IgboRadio and Ka O

.dI

.Taa 7634/ 1090/ 2181 7.5 344,095

Kinyarwanda (kin) NC / Bantu East 10M IGIHE, Rwanda 7825/ 1118/ 2235 12.6 245,933

Luganda (lug) NC / Bantu East 7M MAFAND-MT (Adelani et al.,2022) 4942/ 706/ 1412 15.6 120,119

Luo (luo) Nilo-Saharan East 4M MAFAND-MT (Adelani et al.,2022) 5161/ 737/ 1474 11.7 229,927

Mossi (mos) NC / Gur West 8M MAFAND-MT (Adelani et al.,2022) 4532/ 648/ 1294 9.2 168,141

Naija (pcm) English-Creole West 75M MAFAND-MT (Adelani et al.,2022) 5646/ 806/ 1613 9.4 206,404

Chichewa (nya) NC / Bantu South-East 14M Nation Online Malawi 6250/ 893/ 1785 9.3 263,622

chiShona (sna) NC / Bantu South 12M VOA Shona 6207/ 887/ 1773 16.2 195,834

Kiswahili (swa) NC / Bantu East & Central 98M VOA Swahili 6593/ 942/ 1883 12.7 251,678

Setswana (tsn) NC / Bantu South 14M MAFAND-MT (Adelani et al.,2022) 3489/ 499/ 996 8.8 141,069

Akan/Twi (twi) NC / Kwa West 9M MAFAND-MT (Adelani et al.,2022) 4240/ 605/ 1211 6.3 155,985

Wolof (wol) NC / Senegambia West 5M MAFAND-MT (Adelani et al.,2022) 4593/ 656/ 1312 7.4 181,048

isiXhosa (xho) NC / Bantu South 9M Isolezwe Newspaper 5718/ 817/ 1633 15.1 127,222

Yorùbá (yor) NC / Volta-Niger West 42M Voice of Nigeria and Asejere 6877/ 983/ 1964 11.4 244,144

isiZulu (zul) NC / Bantu South 27M Isolezwe Newspaper 5848/ 836/ 1670 11.0 128,658

Table 1: Languages and Data Splits for MasakhaNER 2.0 Corpus. Language, family (NC: Niger-Congo),

number of speakers, news source, and data split in number of sentences

the Western, Eastern, Central and Southern regions

of Africa. The selected languages cover four lan-

guage families. 17 languages belong to the Niger-

Congo language family, and one language belongs

to each of the Afro-Asiatic (Hausa), Nilo-Saharan

(Luo), and English Creole (Naija) families. Al-

though many languages belong to the Niger-Congo

language family, they have different linguistic char-

acteristics. For instance, Bantu languages (eight in

our selection) make extensive use of afﬁxes, unlike

many languages of non-Bantu subgroups such as

Gur, Kwa, and Volta-Niger.

3.2 Language Characteristics

Script and Word Order

African languages

mainly employ four major writing scripts: Latin,

Arabic, N’ko and Ge’ez. Our focus languages

mostly make use of the Latin script. While N’ko

is still actively used by the Mande languages like

Bambara, the most widely used writing script for

the language is Latin. However, some languages

use additional letters that go beyond the standard

Latin script, e.g., “

”, “

”, and more than

one character letters like “bv”, “gb”, “mpf”, “ntsh”.

17 of the languages are tonal except for Naija,

Kiswahili and Wolof. Nine of the languages make

use of diacritics (e.g., é, ë, ñ). All languages use

the SVO word order, while Bambara additionally

uses the SOV word order.

Morphology and Noun classes

Many African

languages are morphologically rich. According to

the World Atlas of Language Structures (WALS;

Nichols and Bickel,2013), 16 of our languages

employ strong preﬁxing or sufﬁxing inﬂections.

Niger-Congo languages are known for their system

of noun classiﬁcation. 12 of the languages actively

make use of between 6–20 noun classes, includ-

ing all Bantu languages, Ghomálá’, Mossi, Akan

and Wolof (Nurse and Philippson,2006;Payne

et al.,2017;Bodomo and Marfo,2002;Babou and

Loporcaro,2016). While noun classes are often

marked using afﬁxes on the head word in Bantu

languages, some non-Bantu languages, e.g., Wolof

make use of a dependent such as a determiner that

is not attached to the head word. For the other

Niger-Congo languages such as Fon, Ewe, Igbo

and Yorùbá, the use of noun classes is merely ves-

tigial (Konoshenko and Shavarina,2019). Three

of our languages from the Southern Bantu family

(chiShona, isiXhosa and isiZulu) capitalize proper

names after the noun class preﬁx as in the lan-

guage names themselves. This characteristic may

limit transfer from languages without this feature

as NER models overﬁt on capitalization (Mayhew

et al.,2019). Appendix B provides more details

regarding the languages’ linguistic characteristics.

4 MasakhaNER 2.0 Corpus

4.1 Data source and collection

We annotate news articles from local sources. The

choice of the news domain is based on the avail-

ability of data for many African languages and the

variety of named entities types (e.g., person names

and locations) as illustrated by popular datasets

such as CoNLL-03 (Tjong Kim Sang and De Meul-

der,2003).

Table 1 shows the sources and sizes

We also considered using Wikipedia as our data source,

but did not due to quality issues (Alabi et al.,2020).

of the data we use for annotation. Overall, we col-

lected between 4.8K–11K sentences per language

from either a monolingual or a translation corpus.

Monolingual corpus

We collect a large monolin-

gual corpus for nine languages, mostly from local

news articles except for chiShona and Kiswahili

texts, which were crawled from Voice of America

(VOA) websites.

As Yorùbá text was missing dia-

critics, we asked native speakers to manually add

diacritics before annotation. During data collection,

we ensured that the articles are from a variety of

topics e.g. politics, sports, culture, technology, so-

ciety, and education. In total, we collected between

8K–11K sentences per language.

Translation corpus

For the remaining languages

for which we were unable to obtain sufﬁcient

amounts of monolingual data, we use a transla-

tion corpus, MAFAND-MT (Adelani et al.,2022),

which consists of French and English news articles

translated into 11 languages. We note that transla-

tionese may lead to undesired properties, e.g., un-

naturalness. However, we did not observe serious

issues during the annotation. The number of sen-

tences is constrained by the size of the MAFAND-

MT corpus, which is between 4,800–8,000.

4.2 NER Annotation Methodology

We annotated the collected monolingual texts with

the ELISA annotation tool (Lin et al.,2018) with

four entity types: Personal name (

PER

), Loca-

tion (

LOC

), Organization (

ORG

), and date and time

(

DATE

), similar to MasakhaNER 1.0 (Adelani et al.,

2021b). We made use of the MUC-6 annotation

guide.

The annotation was carried out by three na-

tive speakers per language recruited from AI/NLP

communities in Africa. To ensure high-quality an-

notation, we recruited a language coordinator to su-

pervise annotation in each language. We organized

two online workshops to train language coordina-

tors on the NER annotation. As part of the training,

each coordinator annotated 100 English sentences,

which were veriﬁed. Each coordinator then trained

three annotators in their team using both English

and African language texts with the support of the

workshop organizers. All annotators and language

coordinators received appropriate remuneration.6

At the end of annotation, language coordinators

worked with their team to resolve disagreements

4www.voashona.com/ and www.voaswahili.com/

5https://cs.nyu.edu/~grishman/muc6.html

6$10 per hour, annotating about 200 sentences per hour.

Fleiss’ QC ﬂags Fleiss’ QC ﬂags

Lang. Kappa ﬁxed? Lang. Kappa ﬁxed?

bam 0.980 7pcm 0.966 7

bbj 1.000 3nya 0.988 3

ewe 0.991 3sna 0.957 3

fon 0.941 7swa 0.974 3

hau 0.950 7tsn 0.962 7

ibo 0.965 7twi 0.932 7

kin 0.943 7wol 0.979 3

lug 0.950 3xho 0.945 3

luo 0.907 7yor 0.950 3

mos 0.927 7zul 0.953 3

Table 2: Inter-annotator agreement for our datasets cal-

culated using Fleiss’ kappa κat the entity level before

adjudication. QC ﬂags (3) are the languages that ﬁxed

the annotations for all Quality Control ﬂagged tokens.

using the adjudication function of ELISA, which

ensures a high inter-annotator agreement score.

4.3 Quality Control

As discussed in subsection 4.2, language coordi-

nators helped resolve several disagreements in an-

notation prior to quality control. Table 2 reports

the Fleiss Kappa score after the intervention of lan-

guage coordinators (i.e. post-intervention score).

The pre-intervention Fleiss Kappa score was much

lower. For example, for

pcm

, the pre-intervention

Fleiss Kappa score was 0.648 and improved to

0.966 after the language coordinator discussed the

disagreements with the annotators.

For the quality control, annotations were auto-

matically adjudicated when there was agreement,

but were ﬂagged for further review when anno-

tators disagreed on mention spans or types. The

process for reviewing and ﬁxing quality control

issues was voluntary and so not all languages were

further reviewed (see Table 2).

We automatically identiﬁed positions in the an-

notation that were more likely to be annotation

errors and ﬂagged them for further review and cor-

rection. The automatic process ﬂags tokens that are

commonly annotated as a named entity but were

not marked as a named entity in a speciﬁc position.

For example, the token Province may appear com-

monly as part of a named entity and infrequently

not as a named entity, so when it is seen as not

marked it was ﬂagged. Similarly, we ﬂagged to-

kens that had near-zero entropy with regard to a

certain entity type, for example a token almost al-

ways annotated as ORG but very rarely annotated

as PER. We also ﬂagged potential sentence bound-

ary errors by identifying sentences with few tokens

PLM # Lang. Languages in MasakhaNER 2.0

mBERT-cased (110M) 104 swa,yor

XLM-R-base/large

(270M / 550M)

100 hau,swa,xho

mDeBERTaV3 (276M)

100 hau,swa,xho

RemBERT (575M) 110 hau

ibo

nya

sna

swa

xho

yor

zul

AfriBERTa (126M) 11 hau,ibo,kin,pcm,swa,yor

AfroXLMR-base/large

(270M/550M)

20 hau

ibo

kin

nya

pcm

sna

swa

xho,yor,zul

Table 3: Language coverage and size for PLMs.

or sentences which end in a token that appears to

be an abbreviation or acronym. As shown in Table

2, before further adjudication and correction there

was already relatively high inter-annotator agree-

ment measured by Fleiss’ Kappa at the mention

level.

After quality control, we divided the annotation

into training, development, and test splits consist-

ing of 70%, 10%, and 20% of the data respectively.

Appendix A provide details on the number of to-

kens per entity (PER, LOC, ORG, and DATE) and

the fraction of entities in the tokens.

5 Baseline Experiments

5.1 Baseline Models

As baselines, we ﬁne-tune several multilingual

PLMs including mBERT (Devlin et al.,2019b),

XLM-R (base & large; Conneau et al.,2020), mDe-

BERTaV3 (He et al.,2021), AfriBERTa (Ogueji

et al.,2021), RemBERT (Chung et al.,2021), and

AfroXLM-R (base & large; Alabi et al.,2022). We

ﬁne-tune the PLMs on each language’s training

data and evaluate performance on the test set using

HuggingFace Transformers (Wolf et al.,2020).

Massively multilingual PLMs

Table 3 shows

the language coverage and size of different mas-

sively multilingual PLMs trained on 100–110 lan-

guages. mBERT was pre-trained using masked

language modeling (MLM) and next-sentence pre-

diction on 104 languages, including

swa

and

yor

RemBERT was trained with a similar objective, but

makes use of a larger output embedding size dur-

ing pre-training and covers more African languages.

XLM-R was trained only with MLM on 100 lan-

guages and on a larger pre-training corpus. mDe-

BERTaV3 makes use of ELECTRA-style (Clark

et al.,2020) pre-training, i.e., a replaced token de-

tection (RTD) objective instead of MLM.

Africa-centric multilingual PLMs

We also ob-

tained NER models by ﬁne-tuning two PLMs

that are pre-trained on African languages. AfriB-

ERTa (Ogueji et al.,2021) was pre-trained on

less than 1 GB of text covering 11 African lan-

guages, including six of our focus languages, and

has shown impressive performance on NER and

sentiment classiﬁcation for languages in its pre-

training data (Adelani et al.,2021b;Muhammad

et al.,2022). AfroXLM-R (Alabi et al.,2022) is

a language-adapted (Pfeiffer et al.,2020) version

of XLM-R that was ﬁne-tuned on 17 African lan-

guages and three high-resource languages widely

spoken in Africa (“eng”, “fra”, and “ara”). Ap-

pendix J provides the model hyper-parameters for

ﬁne-tuning the PLMs.

5.2 Baseline Results

Table 4 shows the results of training NER models

on each language using the eight multilingual and

Africa-centric PLMs. All PLMs provided good

performance in general. However, we observed

worse results for mBERT and AfriBERTa espe-

cially for languages they were not pre-trained on.

For instance, both models performed between 6–

12 F1 worse for

bbj

wol

zul

compared to

XLM-R-base. We hypothesize that the perfor-

mance drop is largely due to the small number of

African languages covered by mBERT as well as

AfriBERTa’s comparatively small model capacity.

XLM-R-base gave much better performance (

>1.0

F1) on average compared to mBERT and AfriB-

ERTa. We found the larger variants of mBERT

and XLM-R, i.e., RemBERT and XLM-R-large to

give much better performance (

>2.0

F1) than the

smaller models. Their larger capacity facilitates

positive transfer, yielding better performance for

unseen languages. Surprisingly, mDeBERTaV3

provided slightly better results than XLM-R-large

and RemBERT despite its smaller size, demon-

strating the beneﬁts of the RTD pre-training (Clark

et al.,2020).

The best PLM is AfroXLM-R-large, which out-

performs mDeBERTaV3, RemBERT and AfriB-

ERTa by

+1.3

F1,

+2.0

F1 and

+4.0

F1 respec-

tively. Even the performance of its smaller variant,

AfroXLM-R-base is comparable to mDeBERTaV3.

Overall, our baseline results highlight that large

PLMs, PLM with improved pre-training objectives,

and PLMs pre-trained on the target African lan-

guages are able to achieve reasonable baseline per-

formance. Combining these criteria provides im-

proved performance, such as AfroXLM-R-large, a

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

MasakhaNER2.0:Africa-centricTransferLearningforNamedEntityRecognitionDavidIfeoluwaAdelani1;2;,GrahamNeubig3,SebastianRuder4,ShrutiRijhwani3,MichaelBeukman5,ChesterPalen-Michel6,ConstantineLignos6,JesujobaO.Alabi1,ShamsuddeenH.Muhammad7,PeterNabende8,CheikhM.BambaDione9,AndiswaBukula10,Roowei...

展开>> 收起<<

MasakhaNER 2.0 Africa-centric Transfer Learning for Named Entity Recognition David Ifeoluwa Adelani12 Graham Neubig3 Sebastian Ruder4 Shruti Rijhwani3.pdf

共22页,预览5页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

MasakhaNER 2.0 Africa-centric Transfer Learning for Named Entity Recognition David Ifeoluwa Adelani12 Graham Neubig3 Sebastian Ruder4 Shruti Rijhwani3

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: