Detecting Toxic Language: Ontology and BERT-based Approaches for Bulgarian Text

TL;DR

This paper introduces ontology and BERT-based methods for detecting toxic language in Bulgarian, achieving high accuracy and enabling nuanced moderation that preserves valuable information.

Contribution

It presents a novel Bulgarian toxicity ontology, a labeled dataset, and a BERT-based classifier with high F1 score for improved content moderation.

Findings

BERT-based model achieved 0.89 F1 macro score.

Developed a dataset of 4,384 annotated Bulgarian sentences.

Proposed ontology models toxic words in Bulgarian language.

Abstract

Toxic content detection in online communication remains a significant challenge, with current solutions often inadvertently blocking valuable information, including medical terms and text related to minority groups. This paper presents a more nu-anced approach to identifying toxicity in Bulgarian text while preserving access to essential information. The research explores two distinct methodologies for detecting toxic content. The developed methodologies have po-tential applications across diverse online platforms and content moderation systems. First, we propose an ontology that models the potentially toxic words in Bulgarian language. Then, we compose a dataset that comprises 4,384 manually anno-tated sentences from Bulgarian online forums across four categories: toxic language, medical terminology, non-toxic lan-guage, and terms related to minority communities. We then train a BERT-based model for toxic language classification, which reaches a 0.89 F1 macro score. The trained model is directly applicable in a real environment and can be integrated as a com-ponent of toxic content detection systems.