Published Paper Details:

In the era of digital communication, social media platforms have become vital, facilitating global connectivity. However, they also confront severe challenges, such as the proliferation of hate speech and offensive content. Hate speech poses substantial risks by targeting individuals or groups based on attributes like religion, race, or gender. Addressing this issue necessitates a combination of technological, community-driven, and policy-oriented solutions. Due to the lengthy posts, it would be beneficial to identify the specific span of text containing hateful content to assist site moderators with removing hate speech. This paper investigates the application of machine learning and deep learning techniques, specifically Bidirectional Long Short-Term Memory (BiLSTM) networks, for automating the identification of hate speech and offensive content. Using BiLSTM in conjunction with GloVe embedding's, we achieved a considerable F1-score of hate span detection, marking a significant improvement over previous methods. Our automated system represents a critical step toward a safer and more inclusive online environment. By seamlessly integrating machine learning and deep learning, it enhances our ability to detect and combat hate speech effectively. This research underscores the importance of technological advancements in addressing complex social challenges within the digital realm. While significant progress has been made, ongoing efforts in technological innovation, community engagement, and policy development are necessary to create a more respectful and harmonious online space.