Automatic Short Answer Grading onHigh School’s E-Learning Using Semantic Similarity Methods

Grading students’ answers has always been a daunting task which takes lot of teachers’ time. The aim this study is to grade automatically in high school’s e-learning system. grading process must be fast, and the result as close possible teacher assigned grades. We collected total 840 from 40 students for study, each already graded by their teachers. used Python library sentence-transformers three its latest pre-trained machine learning models (all-mpnet-base-v2, all-distilroberta-v1, all-MiniLM-L6-v2) sentence embeddings. Computer grades were calculated using Cosine Similarity. These then compared with both Mean Absolute Error Root Square Error. Our results showed that all-MiniLM-L6-v2 gave most similar had fastest processing Further may include testing these on more students, also fine tune school materials.