Based on the literature on the topic of blended learning collected by CNKI from 2000 to 2021, we analyzed the number distribution of blended learning literature, core authors, research institutions, and further conducted keyword co‐occurrence analysis, emergence detection analysis, and cluster time domain mapping analysis by using the knowledge mapping tool CiteSpace. The study found that the hot spots of domestic blended learning include deep learning, teaching models, neural networks, online learning, blended learning, flipped classroom, blended learning models, genetic algorithms, MOOCs, and so on. The research frontiers are divided into three categories: one for theoretical research, two for practical research on teaching and learning, and three for algorithmic research. In general, the theory and practice of blended learning in China need to be deepened. On the one hand, we should further enhance the localization of blended learning research, and while following the international research hotspots and frontiers, we should explore the formation of blended learning theories and application results suitable for China's actual situation by combining the domestic characteristics; on the other hand, we should further enhance the innovation of blended learning research.