<p>This study conducted a comprehensive survey of relay selection strategies in cooperative device-to-device (D2D) communication, focusing on categorizing, comparing, and synthesizing existing methods. It classified the strategies into three primary groups: social-aware, energy-efficient, and AI-based approaches. Social-aware techniques emphasized the role of trust, user interests, and social graphs in relay selection, addressing the reliability and cooperation challenges in user-centric scenarios. Energy-efficient methods, including battery-aware routing, RF energy harvesting, and optimization-based models, aimed to minimize resource consumption while maintaining performance. AI-driven approaches, comprising supervised learning, deep reinforcement learning (DRL), and federated/transfer learning, demonstrated advanced adaptability and scalability in dynamic and dense network environments. The study identified key trade-offs such as computational complexity, real-time applicability, and scalability constraints, and highlighted the growing relevance of hybrid, privacy-aware, and multi-objective solutions. Furthermore, it outlined open challenges in trust modeling, learning scalability, relay control in 6G, and AI interpretability. The findings aimed to guide future research and practical implementation of efficient and intelligent relay strategies for emerging 6G and IoT ecosystems.</p>