Study on learning path optimization based on reinforcement learning and collaborative filtering

With the development of educational informatization, the application of intelligent instructional software in English education has gradually become a focal point of attention. However, traditional teaching methods fail to meet the needs of personalized learning, and there is an urgent need to improve learning efficiency. To address this issue, this study proposes a learning path planning algorithm based on reinforcement learning and an improved collaborative filtering recommendation algorithm. Through the construction of dynamic knowledge graphs and personalized recommendation technology, intelligent optimization of students' learning paths and resources is achieved. Experimental results indicate that the improved learning path planning algorithm can significantly increase students' mastery rate of knowledge points to 78.6%, with a learning efficiency of 4.8% per hour. The recommendation algorithm achieves a maximum F1 score of 91.0%, with a recommendation time optimization ratio of 20.8% and an average student score improvement rate of 27.3%. The research findings validate the effectiveness of intelligent instructional software in personalized learning. In summary, through optimized algorithm design, the study can effectively enhance the level of intelligence in English language teaching software, providing important technical support for the personalized development of future educational systems.