In this paper, an item-based collaborative filtering recommendation algorithm is first introduced to classify similar target users. After that, according to the characteristics of learners, the learning resource recommendation model is constructed based on the personalized recommendation of learners and learning resources to formulate a realistic learning plan. Finally, a real-time interactive intelligent data mining learning platform (DM_Edu) with self-adaptive learning environment is constructed, and the main architecture and module functions of the platform are constructed. The results show that the personalized learning resource recommendation algorithm proposed in this paper has better performance in accuracy, recall and F1-Score, and is an effective and feasible recommendation algorithm.The application results of the DM_Edu platform found that: teacher-student interactions in the theory classes of skill instruction classes present a large proportion of teachers’ questions in classroom speech, and fewer interactions between teachers and students in speech, which are oriented by teachers’ behaviors and the arrangement of classroom sessions. In the theory class of physical culture, teachers were good at receiving and expressing emotions, and conveyed the spirit and value of physical culture through emotional rendering; however, there was less teacher-student interaction, and the interactive atmosphere was not harmonious enough. In the theory class of health education, teachers are good at leading students to actively participate in the class through words, and teacher-student interactions are frequent; however, the class will be delayed resulting in incomplete class.
