The term recommendation system transforms the way for both e-commerce and entertainment landscapes. The exposure of children to various forms of media, including movies, can significantly impact their social and psychological development. Inappropriate content such as offensive language, violence, and sexuality can have negative effects on children’s perceptions, behavior, and emotional well-being. Recommender systems are crucial in managing the overload of big data. A primary challenge lies in effectively understanding user behavior and item characteristics based on their interest and accompanying the suggested videos, if available. Recently, the application of Graph Neural Network (GNN) techniques has gained traction in recommender systems, since the majority of the data is structured in a graph format. GNNs are popular for recommender systems because they excel in learning representations within the graph-like structures inherent to the information they process. In this study, we employed GNN methods as part of a deep learning approach additionally with a content-based recommendation system. We compared the performance of GNN-based methods against other existing methods in recommending educational videos for children on an animated movie dataset. Researchers utilized GNNs to leverage the inherent graph structure in data, enhancing recommendation system precision and effectiveness through optimization and refinement.