Research on the Design of I&P Teaching Resource Recommendation Method Based on AI and Deep Learning Algorithms

In the context of information technology, traditional recommendation methods for I&P teaching resources face challenges such as outdated content, limited delivery methods, and a low degree of personalization, making it difficult to meet students' diverse and precise learning needs. The main goal of this study is to construct an intelligent and personalized I&P teaching resource recommendation system to enhance the accuracy, engagement, and educational effectiveness of I&P education, thereby supporting the development of the “Three-wide Education” system in higher education institutions. This research introduces AI and deep learning algorithms to address the current pain points in the delivery of I&P educational resources and to meet the practical needs of high-quality development in I&P education in the new era. In terms of methodology, this study adopts multiple models including collaborative filtering, neural collaborative filtering, and FCM. It utilizes user profiling techniques to construct personalized feature dimensions, integrating students' learning behaviors, interest preferences, and knowledge mastery to achieve accurate matching and dynamic optimization of resource recommendations. The research results show that, under two dimensions-recommendation of simple and difficult test questions-and at different recommendation ratios (40%, 50%, 60%), the precision, recall, and Fl scores of different models were analyzed and compared. The HECR model outperforms the other two methods across all ratios, demonstrating significant advantages in coverage and matching capabilities. This contributes to building a fast-responding, highly accurate, and timely-updating I&P resource recommendation system. The innovation of this study lies in the organic integration of N CFin deep learning with traditional algorithms, achieving multi-algorithm collaborative optimization in the recommendation system. Moreover, it constructs a highly adaptable multimodal knowledge point tree model, providing a feasible technical path and theoretical reference for the intelligent upgrading of I&P education in universities in the new era.