Research on Intelligent Grading Algorithms for Subjective Questions Based on Transformer and Weakly Supervised Learning

With the rapid development of educational informatization, the development of precise and efficient intelligent scoring technology for subjective questions has become an urgent task. This paper proposes an intelligent scoring algorithm for subjective questions based on Transformers and weakly supervised learning. The algorithm consists of four core modules: data collection, data preprocessing, model training and optimization, and scoring prediction. By combining the semantic understanding capabilities of pre-trained language models with the data utilization efficiency of weakly supervised learning, the algorithm achieves intelligent scoring for subjective questions in the education field. Experimental results demonstrate that the algorithm achieves excellent performance on subjective question datasets, validating its effectiveness and providing new insights and value for intelligent scoring algorithms in the education field.