Toxic Text Detection In Vietnamese Language

Abstract

The rapid growth of online platforms in recent years, such as social In recent years, the online world has seen an explosion of platforms for communication and sharing. Social networks, forums, and countless websites have created a vast and diverse online landscape. This abundance of content, while exciting, has also introduced new challenges, particularly when it comes to protecting children. The ease of access to the internet can expose them to potential risks, such as encountering toxic language and online bullying. Traditional methods of mitigation, like blocking connections or restricting screen time, can be cumbersome and may not be entirely effective. This paper proposes a novel solution that leverages the power of deep learning. By training deep learning models to identify malicious phrases, our models can recognize various forms of inappropriate language, including both sensitive words and seemingly harmless words used with harmful intent. This intelligent filtering system can be implemented on both the server-side and client-side of online platforms, offering a robust layer of protection for users as they navigate the digital world.