<p>This research aims to enhance the performance of vehicular named data networking (VNDN) in dynamic environments by improving cache hit efficiency and minimizing network response delays. The study proposes a content type-aware routing strategy and an integrated content pre-caching strategy based on content popularity prediction to address the challenges of redundant caching and inefficient content delivery in VNDN. A twofold approach is adopted. First, a content type-aware routing strategy is developed to select optimal forwarding paths based on message attributes and destination node information. Second, the unsupervised learning model Latent Dirichlet Allocation is utilized to predict automotive users’ content request preferences dynamically. The topological relationships among devices in the Internet of Vehicles and the predicted user preferences are integrated to estimate content popularity accurately. This information is used to implement a pre-caching strategy (R-pre-cache) that reduces redundant caching and optimizes cache utilization. Simulation experiments demonstrate that the proposed R-pre-cache strategy significantly outperforms existing caching strategies. Key performance improvements include a higher content delivery ratio, reduced latency, minimized retrieval time, and enhanced cache hit ratios. Additionally, the strategy exhibits excellent scalability and durability, making it suitable for dynamic vehicular environments. The proposed content type-aware routing and pre-caching strategies offer a robust solution for improving VNDN performance in dynamic and unpredictable environments. By leveraging content popularity prediction and efficient caching mechanisms, the study provides a scalable and durable framework for enhancing vehicular network communication, making it a viable architecture for diverse content applications in the Internet of Vehicles.</p>