File size: 1,918 Bytes
83d24b2 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 | from __future__ import annotations
from mteb.abstasks import AbsTaskClassification
from mteb.abstasks.TaskMetadata import TaskMetadata
N_SAMPLES = 2048
class TweetEmotionClassification(AbsTaskClassification):
metadata = TaskMetadata(
name="TweetEmotionClassification",
dataset={
"path": "emotone_ar",
"revision": "0ded8ff72cc68cbb7bb5c01b0a9157982b73ddaf",
},
description="A dataset of 10,000 tweets that was created with the aim of covering the most frequently used emotion categories in Arabic tweets.",
reference="https://link.springer.com/chapter/10.1007/978-3-319-77116-8_8",
type="Classification",
category="s2s",
eval_splits=["train"],
eval_langs=["ara-Arab"],
main_score="accuracy",
date=("2014-01-01", "2016-08-31"),
form=["written"],
domains=["Social"],
task_subtypes=["Sentiment/Hate speech"],
license="Not specified",
socioeconomic_status="mixed",
annotations_creators="human-annotated",
dialect=["ara-arab-EG", "ara-arab-LB", "ara-arab-JO", "ara-arab-SA"],
text_creation="found",
bibtex_citation="""
@inproceedings{al2018emotional,
title={Emotional tone detection in arabic tweets},
author={Al-Khatib, Amr and El-Beltagy, Samhaa R},
booktitle={Computational Linguistics and Intelligent Text Processing: 18th International Conference, CICLing 2017, Budapest, Hungary, April 17--23, 2017, Revised Selected Papers, Part II 18},
pages={105--114},
year={2018},
organization={Springer}
}
""",
n_samples={"train": N_SAMPLES},
avg_character_length={"train": 78.8},
)
def dataset_transform(self):
self.dataset = self.dataset.rename_column("tweet", "text")
self.dataset = self.stratified_subsampling(
self.dataset, seed=self.seed, splits=["train"]
)
|