Hate Speech Identification and Categorization on Social Media Using Bi-LSTM: An Information Science Perspective

Abstract

Online social networks empower individuals with limited influence to exert significant control over specific individuals’ lives and exploit the anonymity or social disconnect offered by the Internet to engage in harassment. Women are commonly attacked due to the prevalent existence of sexism in our culture. Efforts to detect misogyny have improved, but its subtle and profound nature makes it challenging to diagnose, indicating that statistical methods may not be enough. This research article explores the use of deep learning techniques for the automatic detection of hate speech against women on Twitter. It offers further insights into the practical issues of automating hate speech detection in social media platforms by utilizing the model’s capacity to grasp linguistic nuances and context. The results highlight the model’s applicability to information science by addressing the expanding need for better retrieval of hazardous content, scalable content moderation, and metadata organization. This work emphasizes content control in the digital ecosystem. The deep learning-based methods discussed improve the retrieval of data connected to hate speech in the context of a digital archive or social media monitoring system, facilitating study in fields including online harassment, policy formation, and social justice campaigning. The findings not only advance the field of natural language processing but also have practical implications for social media platforms, policymakers, and advocacy groups seeking to combat online harassment and foster inclusive digital spaces for women.