Based on subject knowledge mapping, this paper dynamically collects learning data, portrays learners’ learning situations, and accurately regulates the learning process. Personalized learning path recommendations and learning communities are constructed through learner profiling and learning services. Secondly, structural equation modeling was used to hypothesize the three-level elements of the E-GPPE-C model. Finally, 103 college students in smart teaching classes were taken as research subjects, and the utility of the smart teaching model was analyzed separately through the steps of precondition validation and cross-lag model with random intercepts. The results show that the smart teaching model has