Multimodal Learning Data Intelligent Personalized Learning Resource Recommendation System for Web-Based Classrooms

Traditional recommender systems are difficult to fully explore the complex correlation between multimodal data and cannot meet learners' personalized needs. To this end, this paper proposes an intelligent personalized learning resource recommendation system (MDRS) based on multimodal data, which achieves accurate recommendation of learning resources by integrating advanced technologies such as natural language processing, computer vision, and deep learning, and at the same time, by introducing a multimodal data processing mechanism and personalized knowledge graph construction. First, we introduce a multimodal feature embedding module for the model, which utilizes deep neural networks for feature extraction and embedding of data in different modalities, such as text, image, audio, and video, in order to capture the multidimensional characteristics of resources. Subsequently, through the cross-modal fusion and attention mechanism module, the model performs weighted fusion of multimodal features to strengthen the expression of features related to user learning needs. By dynamically adjusting the weights of each modality, the system is able to automatically focus on the features that are most important for learning resource recommendation and eliminate redundant information. This mechanism ensures the system's adaptive fusion of different modalities, allowing multimodal information to be deeply related at the feature level. Next, we design a personalized knowledge graph module to associate users' learning behaviors with resource knowledge points to provide background knowledge support for the recommendation process, which updates the knowledge graph in real time by tracking learners' learning paths, preferences, and knowledge mastery and optimizes the recommended content based on learners' dynamic needs. Finally, through the recommendation optimization module, the recommendation strategy is dynamically adjusted in combination with user feedback to enhance the learning effect. The experimental results show that the proposed recommendation system outperforms traditional methods in terms of prediction accuracy, and can significantly enhance the learning experience and learning efficiency.