Paper Contents
Abstract
Subjective grading of answers is a vital activity in testing students whose free-text answers are graded on relevance, coherence, and correctness. Manual traditional techniques are extremely time-consuming and subject to human errors. In this paper, an ML- and NLP-based method that grades answers subjectively automatically based on semantic meaning analysis, sentence structure analysis, and keyword relevance analysis is introduced. The model is context-based, with the context employing BERT (Bidirectional Encoder Representations from Transformers). The model employs a scoring and classification mechanism for proper evaluation. The model has been tested and trained on several data sets, like question-and-answer sets for learning, with a mean accuracy of 92.4%. The experiments confirm that the suggested approach works in terms of lesser time for evaluation and higher consistency.
Copyright
Copyright © 2025 Sriramula Pradeep. This is an open access article distributed under the Creative Commons Attribution License.
