In recent years, with the rise of artificial intelligence and deep learning technologies, knowledge graphs can bring more efficient and intelligent solutions to different fields. In this paper, we mainly propose the BiLSTMAttention-CRF model and the model of BERT-BiLSTM-Attention for entity recognition and relationship extraction of text data. Entities and relationships can be recognized by these two models. The entities obtained by knowledge fusion are then utilized to complete the construction of the knowledge graph of English learning in higher vocational colleges. The results show that the attention module added to the BiLSTM-CRF model improves the precision rate, recall rate and F1 value by 2.81%, 3.36%, and 3.63, respectively.The introduction of the BiLSTM-Attention model into the BERT layer also improves the effectiveness of the model. It can be found in the knowledge graph of English learning in higher vocational colleges and universities, and its research hotspot is the Internet. The application of knowledge mapping to the practice of English teaching in higher vocational colleges can significantly improve the English performance and learning attitude.
