Tools
Tools
Erol, Gulbahar (2024) The Impact of Deep Learning on Multilingual Toxic Comments. Masters thesis, Dublin, National College of Ireland.
Preview |
PDF (Master of Science)
Download (1MB) | Preview |
Preview |
PDF (Configuration Manual)
Download (915kB) | Preview |
Abstract
This study covers the effectiveness of deep learning models in detecting multilingual toxic comments. With the rise of social media platforms, there has been an increase in the number of cyberbullying, hate speech, and toxic content. This situation can negatively affect the mental health of individuals. In the study, deep learning methods are used to detect toxic comments and reduce their effects. Previous studies by Singh and Chand (2022) were taken as a reference and expanded, and better deep learning methods were applied. In addition to the studies, F1 score values over 80% were obtained in different languages using multilingual datasets. Most of the previous studies were limited to English datasets, and limited research has been done on multilingual datasets. In this study, a multilingual dataset containing 6 different languages was examined and experiments were conducted using three deep learning methods, namely Convolutional Neural Network (CNN), Long Short-Term Memory (LSTM), and Recurrent Neural Network (RNN). The results of the study showed that these models were successful in detecting toxic comments.
Actions (login required)
 |
View Item |