Datasets:
metadata
annotations_creators:
- human-annotated
language:
- ara
- deu
- eng
- fra
- hin
- ita
- por
- spa
license: cc-by-3.0
multilinguality: multilingual
size_categories:
- 10K<n<100K
task_categories:
- text-classification
task_ids:
- sentiment-analysis
- sentiment-scoring
- sentiment-classification
- hate-speech-detection
paperswithcode_id: tweet_sentiment_multilingual
pretty_name: Tweet Sentiment Multilingual
configs:
- config_name: arabic
data_files:
- path: train/arabic.jsonl.gz
split: train
- path: test/arabic.jsonl.gz
split: test
- path: validation/arabic.jsonl.gz
split: validation
- config_name: default
data_files:
- split: test
path: data/test-*
- split: validation
path: data/validation-*
- split: train
path: data/train-*
- config_name: english
data_files:
- path: train/english.jsonl.gz
split: train
- path: test/english.jsonl.gz
split: test
- path: validation/english.jsonl.gz
split: validation
- config_name: french
data_files:
- path: train/french.jsonl.gz
split: train
- path: test/french.jsonl.gz
split: test
- path: validation/french.jsonl.gz
split: validation
- config_name: german
data_files:
- path: train/german.jsonl.gz
split: train
- path: test/german.jsonl.gz
split: test
- path: validation/german.jsonl.gz
split: validation
- config_name: hindi
data_files:
- path: train/hindi.jsonl.gz
split: train
- path: test/hindi.jsonl.gz
split: test
- path: validation/hindi.jsonl.gz
split: validation
- config_name: italian
data_files:
- path: train/italian.jsonl.gz
split: train
- path: test/italian.jsonl.gz
split: test
- path: validation/italian.jsonl.gz
split: validation
- config_name: portuguese
data_files:
- path: train/portuguese.jsonl.gz
split: train
- path: test/portuguese.jsonl.gz
split: test
- path: validation/portuguese.jsonl.gz
split: validation
- config_name: spanish
data_files:
- path: train/spanish.jsonl.gz
split: train
- path: test/spanish.jsonl.gz
split: test
- path: validation/spanish.jsonl.gz
split: validation
dataset_info:
- config_name: default
features:
- name: text
dtype: large_string
- name: label
dtype: large_string
- name: lang
dtype: string
splits:
- name: test
num_bytes: 836804
num_examples: 6878
- name: validation
num_bytes: 319238
num_examples: 2590
- name: train
num_bytes: 1815279
num_examples: 14523
download_size: 1674044
dataset_size: 2971321
- config_name: sentiment
features:
- name: text
dtype: string
- name: label
dtype:
class_label:
names:
'0': negative
'1': neutral
'2': positive
tags:
- mteb
- text
train-eval-index:
- config: sentiment
task: text-classification
task_id: multi_class_classification
splits:
train_split: train
eval_split: test
col_mapping:
text: text
label: target
metrics:
- type: accuracy
name: Accuracy
- type: f1
name: F1 macro
args:
average: macro
- type: f1
name: F1 micro
args:
average: micro
- type: f1
name: F1 weighted
args:
average: weighted
- type: precision
name: Precision macro
args:
average: macro
- type: precision
name: Precision micro
args:
average: micro
- type: precision
name: Precision weighted
args:
average: weighted
- type: recall
name: Recall macro
args:
average: macro
- type: recall
name: Recall micro
args:
average: micro
- type: recall
name: Recall weighted
args:
average: weighted
A multilingual Sentiment Analysis dataset consisting of tweets in 8 different languages.
| Task category | t2c |
| Domains | Social, Written |
| Reference | https://aclanthology.org/2022.lrec-1.27 |
How to evaluate on this task
You can evaluate an embedding model on this dataset using the following code:
import mteb
task = mteb.get_tasks(["TweetSentimentClassification"])
evaluator = mteb.MTEB(task)
model = mteb.get_model(YOUR_MODEL)
evaluator.run(model)
To learn more about how to run models on mteb task check out the GitHub repitory.
Citation
If you use this dataset, please cite the dataset as well as mteb, as this dataset likely includes additional processing as a part of the MMTEB Contribution.
@inproceedings{barbieri-etal-2022-xlm,
abstract = {Language models are ubiquitous in current NLP, and their multilingual capacity has recently attracted considerable attention. However, current analyses have almost exclusively focused on (multilingual variants of) standard benchmarks, and have relied on clean pre-training and task-specific corpora as multilingual signals. In this paper, we introduce XLM-T, a model to train and evaluate multilingual language models in Twitter. In this paper we provide: (1) a new strong multilingual baseline consisting of an XLM-R (Conneau et al. 2020) model pre-trained on millions of tweets in over thirty languages, alongside starter code to subsequently fine-tune on a target task; and (2) a set of unified sentiment analysis Twitter datasets in eight different languages and a XLM-T model trained on this dataset.},
address = {Marseille, France},
author = {Barbieri, Francesco and
Espinosa Anke, Luis and
Camacho-Collados, Jose},
booktitle = {Proceedings of the Thirteenth Language Resources and Evaluation Conference},
month = jun,
pages = {258--266},
publisher = {European Language Resources Association},
title = {{XLM}-{T}: Multilingual Language Models in {T}witter for Sentiment Analysis and Beyond},
url = {https://aclanthology.org/2022.lrec-1.27},
year = {2022},
}
@article{enevoldsen2025mmtebmassivemultilingualtext,
title={MMTEB: Massive Multilingual Text Embedding Benchmark},
author={Kenneth Enevoldsen and Isaac Chung and Imene Kerboua and Márton Kardos and Ashwin Mathur and David Stap and Jay Gala and Wissam Siblini and Dominik Krzemiński and Genta Indra Winata and Saba Sturua and Saiteja Utpala and Mathieu Ciancone and Marion Schaeffer and Gabriel Sequeira and Diganta Misra and Shreeya Dhakal and Jonathan Rystrøm and Roman Solomatin and Ömer Çağatan and Akash Kundu and Martin Bernstorff and Shitao Xiao and Akshita Sukhlecha and Bhavish Pahwa and Rafał Poświata and Kranthi Kiran GV and Shawon Ashraf and Daniel Auras and Björn Plüster and Jan Philipp Harries and Loïc Magne and Isabelle Mohr and Mariya Hendriksen and Dawei Zhu and Hippolyte Gisserot-Boukhlef and Tom Aarsen and Jan Kostkan and Konrad Wojtasik and Taemin Lee and Marek Šuppa and Crystina Zhang and Roberta Rocca and Mohammed Hamdy and Andrianos Michail and John Yang and Manuel Faysse and Aleksei Vatolin and Nandan Thakur and Manan Dey and Dipam Vasani and Pranjal Chitale and Simone Tedeschi and Nguyen Tai and Artem Snegirev and Michael Günther and Mengzhou Xia and Weijia Shi and Xing Han Lù and Jordan Clive and Gayatri Krishnakumar and Anna Maksimova and Silvan Wehrli and Maria Tikhonova and Henil Panchal and Aleksandr Abramov and Malte Ostendorff and Zheng Liu and Simon Clematide and Lester James Miranda and Alena Fenogenova and Guangyu Song and Ruqiya Bin Safi and Wen-Ding Li and Alessia Borghini and Federico Cassano and Hongjin Su and Jimmy Lin and Howard Yen and Lasse Hansen and Sara Hooker and Chenghao Xiao and Vaibhav Adlakha and Orion Weller and Siva Reddy and Niklas Muennighoff},
publisher = {arXiv},
journal={arXiv preprint arXiv:2502.13595},
year={2025},
url={https://arxiv.org/abs/2502.13595},
doi = {10.48550/arXiv.2502.13595},
}
@article{muennighoff2022mteb,
author = {Muennighoff, Niklas and Tazi, Nouamane and Magne, Lo{\"\i}c and Reimers, Nils},
title = {MTEB: Massive Text Embedding Benchmark},
publisher = {arXiv},
journal={arXiv preprint arXiv:2210.07316},
year = {2022}
url = {https://arxiv.org/abs/2210.07316},
doi = {10.48550/ARXIV.2210.07316},
}
Dataset Statistics
Dataset Statistics
The following code contains the descriptive statistics from the task. These can also be obtained using:
import mteb
task = mteb.get_task("TweetSentimentClassification")
desc_stats = task.metadata.descriptive_stats
{
"test": {
"num_samples": 2048,
"number_of_characters": 169117,
"number_texts_intersect_with_train": 0,
"min_text_length": 4,
"average_text_length": 82.57666015625,
"max_text_length": 200,
"unique_text": 2048,
"unique_labels": 3,
"labels": {
"1": {
"count": 688
},
"2": {
"count": 680
},
"0": {
"count": 680
}
},
"hf_subset_descriptive_stats": {
"arabic": {
"num_samples": 256,
"number_of_characters": 21637,
"number_texts_intersect_with_train": 0,
"min_text_length": 14,
"average_text_length": 84.51953125,
"max_text_length": 140,
"unique_text": 256,
"unique_labels": 3,
"labels": {
"1": {
"count": 86
},
"2": {
"count": 85
},
"0": {
"count": 85
}
}
},
"english": {
"num_samples": 256,
"number_of_characters": 23508,
"number_texts_intersect_with_train": 0,
"min_text_length": 17,
"average_text_length": 91.828125,
"max_text_length": 141,
"unique_text": 256,
"unique_labels": 3,
"labels": {
"1": {
"count": 86
},
"2": {
"count": 85
},
"0": {
"count": 85
}
}
},
"german": {
"num_samples": 256,
"number_of_characters": 19069,
"number_texts_intersect_with_train": 0,
"min_text_length": 9,
"average_text_length": 74.48828125,
"max_text_length": 142,
"unique_text": 256,
"unique_labels": 3,
"labels": {
"1": {
"count": 86
},
"2": {
"count": 85
},
"0": {
"count": 85
}
}
},
"french": {
"num_samples": 256,
"number_of_characters": 24130,
"number_texts_intersect_with_train": 0,
"min_text_length": 23,
"average_text_length": 94.2578125,
"max_text_length": 140,
"unique_text": 256,
"unique_labels": 3,
"labels": {
"1": {
"count": 86
},
"2": {
"count": 85
},
"0": {
"count": 85
}
}
},
"italian": {
"num_samples": 256,
"number_of_characters": 23564,
"number_texts_intersect_with_train": 0,
"min_text_length": 14,
"average_text_length": 92.046875,
"max_text_length": 140,
"unique_text": 256,
"unique_labels": 3,
"labels": {
"1": {
"count": 86
},
"2": {
"count": 85
},
"0": {
"count": 85
}
}
},
"portuguese": {
"num_samples": 256,
"number_of_characters": 18522,
"number_texts_intersect_with_train": 0,
"min_text_length": 24,
"average_text_length": 72.3515625,
"max_text_length": 140,
"unique_text": 256,
"unique_labels": 3,
"labels": {
"1": {
"count": 86
},
"2": {
"count": 85
},
"0": {
"count": 85
}
}
},
"spanish": {
"num_samples": 256,
"number_of_characters": 21014,
"number_texts_intersect_with_train": 0,
"min_text_length": 22,
"average_text_length": 82.0859375,
"max_text_length": 137,
"unique_text": 256,
"unique_labels": 3,
"labels": {
"1": {
"count": 86
},
"2": {
"count": 85
},
"0": {
"count": 85
}
}
},
"hindi": {
"num_samples": 256,
"number_of_characters": 17673,
"number_texts_intersect_with_train": 0,
"min_text_length": 4,
"average_text_length": 69.03515625,
"max_text_length": 200,
"unique_text": 256,
"unique_labels": 3,
"labels": {
"1": {
"count": 86
},
"2": {
"count": 85
},
"0": {
"count": 85
}
}
}
}
},
"train": {
"num_samples": 14712,
"number_of_characters": 1277720,
"number_texts_intersect_with_train": null,
"min_text_length": 2,
"average_text_length": 86.84883088635128,
"max_text_length": 1085,
"unique_text": 14712,
"unique_labels": 3,
"labels": {
"0": {
"count": 4904
},
"1": {
"count": 4904
},
"2": {
"count": 4904
}
},
"hf_subset_descriptive_stats": {
"arabic": {
"num_samples": 1839,
"number_of_characters": 164305,
"number_texts_intersect_with_train": null,
"min_text_length": 11,
"average_text_length": 89.34475258292551,
"max_text_length": 140,
"unique_text": 1839,
"unique_labels": 3,
"labels": {
"0": {
"count": 613
},
"1": {
"count": 613
},
"2": {
"count": 613
}
}
},
"english": {
"num_samples": 1839,
"number_of_characters": 201493,
"number_texts_intersect_with_train": null,
"min_text_length": 29,
"average_text_length": 109.56661228928766,
"max_text_length": 185,
"unique_text": 1839,
"unique_labels": 3,
"labels": {
"0": {
"count": 613
},
"1": {
"count": 613
},
"2": {
"count": 613
}
}
},
"german": {
"num_samples": 1839,
"number_of_characters": 137071,
"number_texts_intersect_with_train": null,
"min_text_length": 7,
"average_text_length": 74.53561718325177,
"max_text_length": 144,
"unique_text": 1839,
"unique_labels": 3,
"labels": {
"0": {
"count": 613
},
"1": {
"count": 613
},
"2": {
"count": 613
}
}
},
"french": {
"num_samples": 1839,
"number_of_characters": 178091,
"number_texts_intersect_with_train": null,
"min_text_length": 16,
"average_text_length": 96.84121805328984,
"max_text_length": 144,
"unique_text": 1839,
"unique_labels": 3,
"labels": {
"0": {
"count": 613
},
"1": {
"count": 613
},
"2": {
"count": 613
}
}
},
"italian": {
"num_samples": 1839,
"number_of_characters": 165828,
"number_texts_intersect_with_train": null,
"min_text_length": 6,
"average_text_length": 90.17292006525285,
"max_text_length": 150,
"unique_text": 1839,
"unique_labels": 3,
"labels": {
"0": {
"count": 613
},
"1": {
"count": 613
},
"2": {
"count": 613
}
}
},
"portuguese": {
"num_samples": 1839,
"number_of_characters": 135761,
"number_texts_intersect_with_train": null,
"min_text_length": 18,
"average_text_length": 73.82327351821642,
"max_text_length": 146,
"unique_text": 1839,
"unique_labels": 3,
"labels": {
"0": {
"count": 613
},
"1": {
"count": 613
},
"2": {
"count": 613
}
}
},
"spanish": {
"num_samples": 1839,
"number_of_characters": 153354,
"number_texts_intersect_with_train": null,
"min_text_length": 19,
"average_text_length": 83.38988580750407,
"max_text_length": 138,
"unique_text": 1839,
"unique_labels": 3,
"labels": {
"0": {
"count": 613
},
"1": {
"count": 613
},
"2": {
"count": 613
}
}
},
"hindi": {
"num_samples": 1839,
"number_of_characters": 141817,
"number_texts_intersect_with_train": null,
"min_text_length": 2,
"average_text_length": 77.11636759108211,
"max_text_length": 1085,
"unique_text": 1839,
"unique_labels": 3,
"labels": {
"0": {
"count": 613
},
"1": {
"count": 613
},
"2": {
"count": 613
}
}
}
}
}
}
This dataset card was automatically generated using MTEB