<p>Social networks, represented as graphs with users as vertices and relationships as edges, are inherently complex and dynamic. Link prediction that forecasts future connections in social networks is vital for platforms like Facebook, WeChat, and LinkedIn. However, as social networks evolve, they exhibit temporal dynamics that pose significant challenges for link prediction, especially when dealing with large-scale networks and a multitude of parameters. Traditional methods often struggle to simultaneously capture both structural and temporal aspects. To overcome these limitations, this paper proposes a novel hybrid model that integrates graph convolutional networks (GCNs) and long short-term memory (LSTM) networks. The GCNs effectively capture the structural information of the network, while the LSTMs model the temporal evolution of connections, improving prediction accuracy. GCNs provide a robust representation of the network's structure, while LSTMs process these representations over time to model temporal dynamics. Experiments conducted on the MUSAE GitHub social network dataset and the KARTAE dataset demonstrate the model's superior performance, achieving accuracies of 92.8% and 98.7%, respectively. Our approach not only simplifies the link prediction process but also outperforms state-of-the-art models, showcasing its potential for broader applications in dynamic social networks. Additionally, the model’s scalability and adaptability to diverse datasets highlight its versatility for real-world applications. Future work will focus on optimizing the model for larger networks, exploring hybrid architectures, and expanding its application to real-time prediction tasks.</p>