Personalized movie recommendation in IoT-enhanced systems using graph convolutional network and multi-layer...

Area of Science:

• Artificial Intelligence
• Computer Science
• Media Studies

Background:

• Cross-cultural communication in animation film distribution faces challenges.
• Integrating emerging technologies like IoT and convolutional networks offers potential solutions.
• Optimizing recommendation systems is crucial for effective content delivery.

Purpose of the Study:

• To optimize communication strategies for animation films in cross-cultural contexts.
• To develop and evaluate a novel movie recommendation model.
• To explore the fusion of IoT and convolutional networks for enhanced film communication.

Main Methods:

• Constructed a collaborative filtering (CF) movie recommendation model using a graph convolutional neural network (GCN).
• Integrated Internet of Things (IoT) and convolutional networks into the communication strategy.
• Validated the GCN-CF model's effectiveness through comparative experiments on MovieLens datasets.
• Introduced attention mechanisms and auxiliary information to further enhance model performance.

Main Results:

• The GCN-CF model achieved the lowest Root Mean Square Error (RMSE) on MovieLens 100K (0.8762) and 1M (0.8275) datasets.
• Demonstrated significant performance improvements over traditional models, with RMSE reductions of 0.6-5.2%.
• Attention mechanisms and auxiliary information improved GCN-CF performance by 0.4%.

Conclusions:

• The GCN-CF model offers superior performance for movie recommendation in cross-cultural contexts.
• The integration of IoT and convolutional networks enhances animation film communication strategies.
• Findings provide novel insights for film production, communication, cultural exchange, and the global media industry.