The Application of Knowledge Graph Convolutional Network-Based Film and Television Interaction Under Artificial Intelligence

Abstract

This study aims to explore the application and innovation of deep learning in film and television interaction in the context of artificial intelligence. By analyzing the interaction needs of users with films and television, the Knowledge Graph Convolutional Network (KGCN) algorithm is introduced, taking into account both user interests and the associated information between films and television. A film recommendation model based on KGCN fused with user interests is proposed. This model extensively explores the relational and semantic information between film works to provide users with richer and more diverse recommendation content, meeting their diverse viewing needs and interactions. Through evaluating the model performance, the proposed model exhibits lower loss function values in terms of convergence. The model reaches a basic stable state with a loss function value of around 0.40 after 43 iterations, far superior to the baseline algorithm Convolutional Neural Network (CNN). In terms of accuracy, the proposed model achieves an Accuracy value of 96.53%, representing an improvement of at least 4.80%. Moreover, the precision and F1 values of the prediction accuracy are improved by over 4%. Additionally, in terms of Area Under the Curve (AUC) value, the proposed model also demonstrates a significant advantage, reaching a level of 98.05%. Therefore, the film recommendation model proposed in this study, which is based on KGCN fused with user interests, possesses significant advantages in film recommendation tasks, providing strong reference and guidance for the improvement and optimization of film recommendation systems.

Keywords

• Deep learning
• KGCN
• knowledge graph
• film and television interaction
• artificial intelligence