Online Social Networks (OSNs) serve as dynamic platforms for a myriad of activities, from e-commerce and collaborations to social networking, fostering extensive information exchange. Despite their immense potential, the open and dispersed nature of OSNs creates a breeding ground for both authentic communication and the dissemination of deceptive content and malicious activities, posing a threat to user trust and overall OSN usage. While platforms aim to deter such incidents, the sheer volume of data overwhelms their processing capacities, compounded by attackers’ adaptive tactics. This article delves into the growing field of studying user trustworthiness within OSNs, offering a comprehensive overview of recent research works. Our focus lies in constructing robust trust and reputation systems, assessing and predicting trust scores by considering user interactions, profiles, preferences, ratings, and feedback. Notably, existing implementations primarily leverage graph-based approaches, grappling with issues like path dependency, trust decay, opinion conflict, and attack resistance. Drawing insights from the literature review, the need for comprehensive tools or methodologies becomes evident. Pinpointing key factors influencing trust scores, accounting for OSN diversity, emerges as crucial. Real-time data handling tools are imperative for dependable trust predictions, urging the integration of such tools into OSNs. While machine learning approaches have been explored, this article advocate further investigation into deep learning and reinforcement learning methods for more effective solutions in this evolving landscape.