Skip to content

Latest commit

 

History

History
77 lines (53 loc) · 4.18 KB

README.md

File metadata and controls

77 lines (53 loc) · 4.18 KB

APEACH - Korean Hate Speech Evaluation Datasets

APEACH is the first crowd-generated Korean evaluation dataset for hate speech detection. Sentences of the dataset are created by anonymous participants using an online crowdsourcing platform DeepNatural AI.

  • Sample Code : base

Download

You can download benchmark set APEACH. APEACH/test.csv in this repository.

Dataset Description

  • APEACH : A hate-speech evaluation dataset generated in 2021, using generation method followd by APEACH paper.

Guidelines

APEACH-GUIDELINE

Topics

Lengths

Paper

Experiment Code

base

Experiment Results

Name Beep! Dev Dataset Apeach (Ours)
SoongsilBERT-Base 0.8261 0.8424
SoongsilBERT-Small 0.8149 0.8228
KcBERT-base 0.8088 0.8086
KcBERT-large 0.8295 0.8116
DistillKoBERT 0.7570 0.7715
KoELECTRA-V3 0.7920 0.8101
KoBERT 0.8030 0.7885

We also share BEST model of our dataset which we trained in this experiment as checkpoint, demo webite and api.

Citation

@inproceedings{yang-etal-2022-apeach,
    title = "{APEACH}: Attacking Pejorative Expressions with Analysis on Crowd-Generated Hate Speech Evaluation Datasets",
    author = "Yang, Kichang  and
      Jang, Wonjun  and
      Cho, Won Ik",
    booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2022",
    month = dec,
    year = "2022",
    address = "Abu Dhabi, United Arab Emirates",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2022.findings-emnlp.525",
    pages = "7076--7086",
    abstract = "In hate speech detection, developing training and evaluation datasets across various domains is the critical issue. Whereas, major approaches crawl social media texts and hire crowd-workers to annotate the data. Following this convention often restricts the scope of pejorative expressions to a single domain lacking generalization. Sometimes domain overlap between training corpus and evaluation set overestimate the prediction performance when pretraining language models on low-data language. To alleviate these problems in Korean, we propose APEACH that asks unspecified users to generate hate speech examples followed by minimal post-labeling. We find that APEACH can collect useful datasets that are less sensitive to the lexical overlaps between the pretraining corpus and the evaluation set, thereby properly measuring the model performance.",
}

Contributors

The main contributors of the work ( * : equal contribution) :

License

Creative Commons License

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.