Hate speech on social media platforms has severe impacts on individuals, online communities, and society. Platforms are criticized for shirking their responsibilities to effectively moderate hate speech on their platforms. However, Various challenges, including implicit expressions, complicate the task of detecting hate speech. Consequently, developing and tuning algorithms for improving the automated detection of hate speech has emerged as a crucial research topic. This paper aims to contribute to this rapidly emerging field by outlining how the adoption of natural language processing and machine learning technologies has helped hate speech detection, delving into the latest mainstream detection techniques and their performance, and offering a comprehensive review of the literature on hate speech detection online including the notable challenges and respective mitigating efforts. This paper proposes the integration of interdisciplinary perspectives into deep learning models to enhance the generalization of models, providing a new agenda for future research.