A System to Correct Toxic Expression with BERT

Motonobu Yoshida, Kazuyuki Matsumoto, Minoru Yoshida, Kenji Kita

2022

Abstract

This paper describes a system for converting posts with toxic expression on social media, such as those containing slander and libel, into less-toxic sentences. In recent years, the number of social media users as well as the cases of online flame wars has been increasing. Therefore, to prevent flaming, we first use a prediction model based on Bidirectional Encoder Representations from Transformers (BERT) to determine whether a sentence is likely to be flamed before it is posted. The highest classification accuracy recorded 82% with the Japanese Spoken Language Field Adaptive BERT Model (Japanese Spoken BERT model) as a pre-trained model. Then, for sentences that are judged to be toxic, we propose a system that uses BERT’s masked word prediction to convert toxic expressions into safe expressions, thereby converting them into sentences with mitigated aggression. In addition, the BERTScore is used to quantify whether the meaning of the converted sentence has changed in meaning compared to the original sentence and evaluate whether the modified sentence is safe while preserving the meaning of the original sentence.

Paper Citation

in Harvard Style

Yoshida M., Matsumoto K., Yoshida M. and Kita K. (2022). A System to Correct Toxic Expression with BERT. In Proceedings of the 14th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management (IC3K 2022) - Volume 2: KEOD; ISBN 978-989-758-614-9, SciTePress, pages 92-97. DOI: 10.5220/0011586100003335

in Bibtex Style

@conference{keod22,
author={Motonobu Yoshida and Kazuyuki Matsumoto and Minoru Yoshida and Kenji Kita},
title={A System to Correct Toxic Expression with BERT},
booktitle={Proceedings of the 14th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management (IC3K 2022) - Volume 2: KEOD},
year={2022},
pages={92-97},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0011586100003335},
isbn={978-989-758-614-9},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 14th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management (IC3K 2022) - Volume 2: KEOD
TI - A System to Correct Toxic Expression with BERT
SN - 978-989-758-614-9
AU - Yoshida M.
AU - Matsumoto K.
AU - Yoshida M.
AU - Kita K.
PY - 2022
SP - 92
EP - 97
DO - 10.5220/0011586100003335
PB - SciTePress