Efficient Toxicity Detection in Gaming Chats: A Comparative Study of Embeddings, Fine-Tuned Transformers and LLMs

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Tereshchenko, Yehor, Hämäläinen, Mika
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866918164575551488
author Tereshchenko, Yehor
Hämäläinen, Mika
author_facet Tereshchenko, Yehor
Hämäläinen, Mika
contents This paper presents a comprehensive comparative analysis of Natural Language Processing (NLP) methods for automated toxicity detection in online gaming chats. Traditional machine learning models with embeddings, large language models (LLMs) with zero-shot and few-shot prompting, fine-tuned transformer models, and retrieval-augmented generation (RAG) approaches are evaluated. The evaluation framework assesses three critical dimensions: classification accuracy, processing speed, and computational costs. A hybrid moderation system architecture is proposed that optimizes human moderator workload through automated detection and incorporates continuous learning mechanisms. The experimental results demonstrate significant performance variations across methods, with fine-tuned DistilBERT achieving optimal accuracy-cost trade-offs. The findings provide empirical evidence for deploying cost-effective, efficient content moderation systems in dynamic online gaming environments.
format Preprint
id arxiv_https___arxiv_org_abs_2510_17924
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Efficient Toxicity Detection in Gaming Chats: A Comparative Study of Embeddings, Fine-Tuned Transformers and LLMs
Tereshchenko, Yehor
Hämäläinen, Mika
Computation and Language
Artificial Intelligence
I.2.7
This paper presents a comprehensive comparative analysis of Natural Language Processing (NLP) methods for automated toxicity detection in online gaming chats. Traditional machine learning models with embeddings, large language models (LLMs) with zero-shot and few-shot prompting, fine-tuned transformer models, and retrieval-augmented generation (RAG) approaches are evaluated. The evaluation framework assesses three critical dimensions: classification accuracy, processing speed, and computational costs. A hybrid moderation system architecture is proposed that optimizes human moderator workload through automated detection and incorporates continuous learning mechanisms. The experimental results demonstrate significant performance variations across methods, with fine-tuned DistilBERT achieving optimal accuracy-cost trade-offs. The findings provide empirical evidence for deploying cost-effective, efficient content moderation systems in dynamic online gaming environments.
title Efficient Toxicity Detection in Gaming Chats: A Comparative Study of Embeddings, Fine-Tuned Transformers and LLMs
topic Computation and Language
Artificial Intelligence
I.2.7
url https://arxiv.org/abs/2510.17924