With the advent of the era of big data, the identification of ways to extract useful knowledge from massive amounts of data has become the key to the development of methods and application research analysis of knowledge graphs. Because the knowledge graph is a type of networked knowledge base with strong expressive ability. Applied in the field of engineering education, it can help students quickly find the learning focus and improve learning efficiency. Therefore, this study proposes the combination of the bidirectional long short-term memory–conditional random fields (BILSTM–CRF) model with the BERT model to build a knowledge graph in the field of engineering education. Given that BILSTM can capture an increased amount of contextual information from the sentence and solve the problem of gradient disappearance during training, the CRF layer can ensure the effectiveness of the final prediction result. The BERT model can fully describe character-level, word-level, sentence-level, and even inter-sentence relationship characteristics. It is considerably suitable for the data in the field of engineering education evaluated in this study. Since the extracted data will always be missing data, this study proposes to use the knowledge reasoning model TransE to supplement and improve the extracted knowledge map data to realize the continuous update of the knowledge map.