Dataset Card for BasqueGLUE

Dataset Summary

Natural Language Understanding (NLU) technology has improved significantly over the last few years, and multitask benchmarks such as GLUE are key to evaluate this improvement in a robust and general way. These benchmarks take into account a wide and diverse set of NLU tasks that require some form of language understanding, beyond the detection of superficial, textual clues. However, they are costly to develop and language-dependent, and therefore they are only available for a small number of languages.

We present BasqueGLUE, the first NLU benchmark for Basque, which has been elaborated from previously existing datasets and following similar criteria to those used for the construction of GLUE and SuperGLUE. BasqueGLUE is freely available under an open license.

Dataset	|Train|	|Val|	|Test|	Task	Metric	Domain
NERCid	51,539	12,936	35,855	NERC	F1	News
NERCood	64,475	14,945	14,462	NERC	F1	News, Wikipedia
FMTODeu_intent	3,418	1,904	1,087	Intent classification	F1	Dialog system
FMTODeu_slot	19,652	10,791	5,633	Slot filling	F1	Dialog system
BHTCv2	8,585	1,857	1,854	Topic classification	F1	News
BEC2016eu	6,078	1,302	1,302	Sentiment analysis	F1	Twitter
VaxxStance	864	206	312	Stance detection	MF1*	Twitter
QNLIeu	1,764	230	238	QA/NLI	Acc	Wikipedia
WiCeu	408,559	600	1,400	WSD	Acc	Wordnet
EpecKorrefBin	986	320	587	Coreference resolution	Acc	News

Supported Tasks and Leaderboards

This benchmark comprises the following tasks:

NERCid

This dataset contains sentences from the news domain with manually annotated named entities. The data is the merge of EIEC (a dataset of a collection of news wire articles from Euskaldunon Egunkaria newspaper, (Alegria et al. 2004)), and newly annotated data from naiz.eus. The data is annotated following the BIO annotation scheme over four categories: person, organization, location, and miscellaneous.

NERCood

This dataset contains sentences with manually annotated named entities. The training data is the merge of EIEC (a dataset of a collection of news wire articles from Euskaldunon Egunkaria newspaper, (Alegria et al. 2004)), and newly annotated data from naiz.eus. The data is annotated following the BIO annotation scheme over four categories: person, organization, location, and miscellaneous. For validation and test sets, sentences from Wikipedia were annotated following the same annotation guidelines.

FMTODeu_intent

This dataset contains utterance texts and intent annotations drawn from the manually-annotated Facebook Multilingual Task Oriented Dataset (FMTOD) (Schuster et al. 2019). Basque translated data was drawn from the datasets created for Building a Task-oriented Dialog System for languages with no training data: the Case for Basque (de Lacalle et al. 2020). The examples are annotated with one of 12 different intent classes corresponding to alarm, reminder or weather related actions.

FMTODeu_slot

This dataset contains utterance texts and sequence intent argument annotations designed for slot filling tasks, drawn from the manually-annotated Facebook Multilingual Task Oriented Dataset (FMTOD) (Schuster et al. 2019). Basque translated data was drawn from the datasets created for Building a Task-oriented Dialog System for languages with no training data: the Case for Basque (de Lacalle et al. 2020). The task is a sequence labelling task similar to NERC, following BIO annotation scheme over 11 categories.

BHTCv2

The corpus contains 12,296 news headlines (brief article descriptions) from the Basque weekly newspaper Argia. Topics are classified uniquely according to twelve thematic categories.

BEC2016eu

The Basque Election Campaign 2016 Opinion Dataset (BEC2016eu) is a new dataset for the task of sentiment analysis, a sequence classification task, which contains tweets about the campaign for the Basque elections from 2016. The crawling was carried out during the election campaign period (2016/09/09-2016/09/23), by monitoring the main parties and their respective candidates. The tweets were manually annotated as positive, negative or neutral.

VaxxStance

The VaxxStance (Agerri et al., 2021) dataset originally provides texts and stance annotations for social media texts around the anti-vaccine movement. Texts are given a label indicating whether they express an AGAINST, FAVOR or NEUTRAL stance towards the topic.

QNLIeu

This task includes the QA dataset ElkarHizketak (Otegi et al. 2020), a low resource conversational Question Answering (QA) dataset for Basque created by native speaker volunteers. The dataset is built on top of Wikipedia sections about popular people and organizations, and it contains around 400 dialogues and 1600 question and answer pairs. The task was adapted into a sentence-pair binary classification task, following the design of QNLI for English (Wang et al. 2019). Each question and answer pair are given a label indicating whether the answer is entailed by the question.

WiCeu

Word in Context or WiC (Pilehvar and Camacho-Collados 2019) is a word sense disambiguation (WSD) task, designed as a particular form of sentence pair binary classification. Given two text snippets and a polyse mous word that appears in both of them (the span of the word is marked in both snippets), the task is to determine whether the word has the same sense in both sentences. This dataset is based on the EPEC-EuSemcor (Pociello et al. 2011) sense-tagged corpus.

EpecKorrefBin

EPEC-KORREF-Bin is a dataset derived from EPEC-KORREF (Soraluze et al. 2012), a corpus of Basque news documents with manually annotated mentions and coreference chains, which we have been converted into a binary classification task. In this task, the model has to predict whether two mentions from a text, which can be pronouns, nouns or noun phrases, are referring to the same entity.

Leaderboard

Results obtained for two BERT base models as a baseline for the Benchmark.

	AVG	NERC	F_intent	F_slot	BHTC	BEC	Vaxx	QNLI	WiC	coref
Model		F1	F1	F1	F1	F1	MF1	acc	acc	acc
BERTeus	73.23	81.92	82.52	74.34	78.26	69.43	59.30	74.26	70.71	68.31
ElhBERTeu	73.71	82.30	82.24	75.64	78.05	69.89	63.81	73.84	71.71	65.93

The results obtained on NERC are the average of in-domain and out-of-domain NERC.

Languages

Data are available in Basque (BCP-47 eu)

Dataset Structure

Data Instances

NERCid/NERCood

An example of 'train' looks as follows:

{
  "idx": 0,
  "tags": ["O", "O", "O", "O", "B-ORG", "O", ...],
  "tokens": ["Greba", "orokorrera", "deitu", "du", "EHk", "27rako", ...]
}

FMTODeu_intent

An example of 'train' looks as follows:

{
  "idx": 0,
  "label": "alarm/modify_alarm", 
  "text": "aldatu alarma 7am-tik 7pm-ra , mesedez"
}

FMTODeu_slot

An example of 'train' looks as follows:

{
  "idx": 923, 
  "tags": ["O", "B-reminder/todo", "I-datetime", "I-datetime", "B-reminder/todo"], 
  "tokens": ["gogoratu", "zaborra", "gaur", "gauean", "ateratzea"]
}

BHTCv2

An example of 'test' looks as follows:

{
  "idx": 0, 
  "label": "Gizartea", 
  "text": "Genero berdintasunaz, hezkuntzaz eta klase gizarteaz hamar liburu baino gehiago..."
}

BEC2016eu

An example of 'test' looks as follows:

{
  "idx": 0,
  "label": "NEU",
  "text": '"Emandako hitza bete egingo dut" Urkullu\nBa galdeketa enegarrenez daramazue programan (ta zuen AHTa...)\n#I25debatea #URL"'
}

VaxxStance

An example of 'train' looks as follows:

{
  "idx": 0, 
  "label": "FAVOR", 
  "text": "\"#COVID19 Oraingo datuak, izurriaren dinamika, txertoaren eragina eta birusaren..
}

QNLIeu

An example of 'train' looks as follows:

{
  "idx": 1, 
  "label": "not_entailment", 
  "question": "Zein posiziotan jokatzen du Busquets-ek?", 
  "sentence": "Busquets 23 partidatan izan zen konbokatua eta 2 gol sartu zituen."
}

WiCeu

An example of 'test' looks as follows:

{
  "idx": 16, 
  "label": false, 
  "word": "udal", 
  "sentence1": "1a . Lekeitioko udal mugarteko Alde Historikoa Birgaitzeko Plan Berezia behin...", 
  "sentence2": "Diezek kritikatu egin zuen EAJk zenbait udaletan EH gobernu taldeetatik at utzi...", 
  "start1": 16, 
  "start2": 40, 
  "end1": 21, 
  "end2": 49
}

EpecKorrefBin

An example of 'train' looks as follows:

{
  "idx": 6, 
  "label": false, 
  "text": "Isuntza da faborito nagusia Elantxobeko banderan . ISUNTZA trainerua da faborito nagusia bihar Elantxoben jokatuko den bandera irabazteko .", 
  "span1_text": "Elantxobeko banderan", 
  "span2_text": "ISUNTZA trainerua", 
  "span1_index": 4, 
  "span2_index": 8
  }

Data Fields

NERCid

• tokens: a list of string features
• tags: a list of entity labels, with possible values including person (PER), location (LOC), organization (ORG), miscellaneous (MISC)
• idx: an int32 feature

NERCood

• tokens: a list of string features
• tags: a list of entity labels, with possible values including person (PER), location (LOC), organization (ORG), miscellaneous (MISC)
• idx: an int32 feature

FMTODeu_intent

• text: a string feature
• label: an intent label, with possible values including:
  • alarm/cancel_alarm
  • alarm/modify_alarm
  • alarm/set_alarm
  • alarm/show_alarms
  • alarm/snooze_alarm
  • alarm/time_left_on_alarm
  • reminder/cancel_reminder
  • reminder/set_reminder
  • reminder/show_reminders
  • weather/checkSunrise
  • weather/checkSunset
  • weather/find
• idx: an int32 feature

FMTODeu_slot

• tokens: a list of string features
• tags: a list of intent labels, with possible values including:
  • datetime
  • location
  • negation
  • alarm/alarm_modifier
  • alarm/recurring_period
  • reminder/noun
  • reminder/todo
  • reminder/reference
  • reminder/recurring_period
  • weather/attribute
  • weather/noun
• idx: an int32 feature

BHTCv2

• text: a string feature
• label: a polarity label, with possible values including neutral (NEU), negative (N), positive (P)
• idx: an int32 feature

BEC2016eu

• text: a string feature
• label: a topic label, with possible values including:
  • Ekonomia
  • Euskal Herria
  • Euskara
  • Gizartea
  • Historia
  • Ingurumena
  • Iritzia
  • Komunikazioa
  • Kultura
  • Nazioartea
  • Politika
  • Zientzia
• idx: an int32 feature

VaxxStance

• text: a string feature
• label: a stance label, with possible values including AGAINST, FAVOR, NONE
• idx: an int32 feature

QNLIeu

• question: a string feature
• sentence: a string feature
• label: an entailment label, with possible values including entailment, not_entailment
• idx: an int32 feature

WiCeu

• word: a string feature
• sentence1: a string feature
• sentence2: a string feature
• label: a boolean label indicating sense agreement, with possible values including true, false
• start1: an int feature indicating character position where word occurence begins in first sentence
• start2: an int feature indicating character position where word occurence begins in second sentence
• end1: an int feature indicating character position where word occurence ends in first sentence
• end2: an int feature indicating character position where word occurence ends in second sentence
• idx: an int32 feature

EpecKorrefBin

• text: a string feature.
• label: a boolean coreference label, with possible values including true, false.
• span1_text: a string feature
• span2_text: a string feature
• span1_index: an int feature indicating token index where span1_text feature occurs in text
• span2_index: an int feature indicating token index where span2_text feature occurs in text
• idx: an int32 feature

Data Splits

Dataset	|Train|	|Val|	|Test|
NERCid	51,539	12,936	35,855
NERCood	64,475	14,945	14,462
FMTODeu_intent	3,418	1,904	1,087
FMTODeu_slot	19,652	10,791	5,633
BHTCv2	8,585	1,857	1,854
BEC2016eu	6,078	1,302	1,302
VaxxStance	864	206	312
QNLIeu	1,764	230	238
WiCeu	408,559	600	1,400
EpecKorrefBin	986	320	587

Dataset Creation

Curation Rationale

We believe that BasqueGLUE is a significant contribution towards developing NLU tools in Basque, which we believe will facilitate the technological advance for the Basque language. In order to create BasqueGLUE we took as a reference the GLUE and SuperGLUE frameworks. When possible, we re-used existing datasets for Basque, adapting them to the corresponding task formats if necessary. Additionally, BasqueGLUE also includes six new datasets that have not been published before. In total, BasqueGLUE consists of nine Basque NLU tasks and covers a wide range of tasks with different difficulties across several domains. As with the original GLUE benchmark, the training data for the tasks vary in size, which allows to measure the performance of how the models transfer knowledge across tasks.

Additional Information

Dataset Curators

Gorka Urbizu [1], Iñaki San Vicente [1], Xabier Saralegi [1], Rodrigo Agerri [2] and Aitor Soroa [2]

Affiliation of the authors:

[1] orai NLP Technologies

[2] HiTZ Center - Ixa, University of the Basque Country UPV/EHU

Licensing Information

Each dataset of the BasqueGLUE benchmark has it's own license (due to most of them being or being derived from already existing datasets). See their respective README files for details.

Here we provide a brief summary of their licenses:

Dataset	License
NERCid	CC BY-NC-SA 4.0
NERCood	CC BY-NC-SA 4.0
FMTODeu_intent	CC BY-NC-SA 4.0
FMTODeu_slot	CC BY-NC-SA 4.0
BHTCv2	CC BY-NC-SA 4.0
BEC2016eu	Twitter's license + CC BY-NC-SA 4.0
VaxxStance	Twitter's license + CC BY 4.0
QNLIeu	CC BY-SA 4.0
WiCeu	CC BY-NC-SA 4.0
EpecKorrefBin	CC BY-NC-SA 4.0

For the rest of the files of the benchmark, including the loading and evaluation scripts, the following license applies:

Copyright (C) by Orai NLP Technologies. This benchmark and evaluation scripts are licensed under the Creative Commons Attribution Share Alike 4.0 International License (CC BY-SA 4.0). To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Citation Information

@InProceedings{urbizu2022basqueglue,
  author    = {Urbizu, Gorka  and  San Vicente, Iñaki  and  Saralegi, Xabier  and  Agerri, Rodrigo  and  Soroa, Aitor},
  title     = {BasqueGLUE: A Natural Language Understanding Benchmark for Basque},
  booktitle      = {Proceedings of the Language Resources and Evaluation Conference},
  month          = {June},
  year           = {2022},
  address        = {Marseille, France},
  publisher      = {European Language Resources Association},
  pages     = {1603--1612},
  abstract  = {Natural Language Understanding (NLU) technology has improved significantly over the last few years and multitask benchmarks such as GLUE are key to evaluate this improvement in a robust and general way. These benchmarks take into account a wide and diverse set of NLU tasks that require some form of language understanding, beyond the detection of superficial, textual clues. However, they are costly to develop and language-dependent, and therefore they are only available for a small number of languages. In this paper, we present BasqueGLUE, the first NLU benchmark for Basque, a less-resourced language, which has been elaborated from previously existing datasets and following similar criteria to those used for the construction of GLUE and SuperGLUE. We also report the evaluation of two state-of-the-art language models for Basque on BasqueGLUE, thus providing a strong baseline to compare upon. BasqueGLUE is freely available under an open license.},
  url       = {https://aclanthology.org/2022.lrec-1.172}
}

Contributions

Thanks to @richplant for adding this dataset to hugginface.