Impact of a Knowledge Graph Technology Supported English Terminology Learning System on Mastery of English in Disciplines

Abstract

In the face of the challenge of fragmented knowledge in English, the study starts from helping students master terminology knowledge in a more systematic way, and constructs an English terminology learning system driven by knowledge graph technology. By combining top-down and bottom-up approaches, the study utilizes the pre-trained BERT-BiLSTM-CRF model to extract and construct a knowledge graph for the subject of English at university. The system integrates key technologies such as efficient resource indexing, online collaborative text-based cross-domain mapping extension, and time-ordered knowledge inference. Firstly, 32703 knowledge point entities are mined from 20 core courses, and the accuracy of entity alignment can reach up to 95.38%, while the average F1 value of recognizing various types of entities is also stable at about 94.40%. In actual teaching, the experimental class using this system constructed in this paper, the performance improvement is very obvious - the average score directly rose from 68 to 86, to a certain extent, narrowing the achievement gap between students, and there is no low-scoring students. At the same time students were more active in the classroom, with the level of assessment of learning increasing from 3.58 to 4.77, and the intention to learn behavior increasing dramatically from 2.87 to 4.56. Satisfaction with the system and instructional design even reached a high score of 4.78 and 4.64 respectively. Regression analysis shows that the system itself explains 8.1% to 14.5% of the improvement in each English proficiency.