Geometry Arithmetic Problem Recommendation Based on Scene-Enhanced BERT

Intelligent recommendation plays an important role in the field of intelligent education, among which arithmetic problem recommendation is an important research field. Geometry arithmetic problems are a major category of arithmetic problems which play a crucial role in nurturing students’ spatial perception, logical thinking, and computational skills. This paper proposes a scene-enhanced BERT model for recommending geometry arithmetic problems. Firstly, a multi-layer classification task of geometric knowledge was constructed for fine-tuning the BERT model. Secondly, this BERT model is used to vectorize the problem to be queried and the problem text in the database. Finally, the query questions are matched with questions under the same knowledge points using vector similarity calculations with diversity correction and recommended according to the score. Experimental results demonstrate that our proposed method has achieved good efficiency and accuracy in recommending geometric arithmetic problems. Overall, this research offers an effective recommendation approach for the geometry arithmetic problem task, providing valuable insights for intelligent education field and tutoring system.