A machine learning model capable of detecting toxic social media comments with great accuracy — thus paving the way for safer digital interactions — has been developed by a team of computer scientists from East West University in Bangladesh and the University of South Australia.

According to the computer scientists, the model is reportedly 87% accurate in classifying toxic and non-toxic text without the use of manual identification — an improvement over existing automated detection systems, many of which produce false positives.

"Despite efforts by social media platforms to limit toxic content, manually identifying harmful comments is impractical due to the sheer volume of online interactions, with 5.56 billion internet users in the world today," the researchers explained. “Removing toxic comments from online network platforms is vital to curbing the escalating abuse and ensuring respectful interactions in the social media space."

To develop the model, the team tested three machine learning models on a dataset of English and Bangla comments gathered from social media platforms like Facebook, YouTube and Instagram.

The team’s algorithm reportedly achieved a rate of accuracy of roughly 87.6%, thus outperforming other models that achieved accuracy rates of 69.9% and 83.4%.

The team expects to improve the model by incorporating deep learning techniques and expanding the dataset to include additional languages as well as regional dialects.

The study, “An Approach to Text Toxicity Classification Using Machine Learning in Native Language,” appears in the journal 2024 International Conference on Innovation and Intelligence for Informatics, Computing, and Technologies.