Intelligent Scoring System for English Subjective Questions Based on Neural Networks

An intelligent scoring system for English subjective questions was created using neural networks in this study. The created system automates the grading process to provide efficient and accurate evaluations for essays and textual responses. Various layers of neural network architecture such as embedding, convolutional, long short-term memory (LSTM) network, intrinsic feature, and fully connected layers are included in the system. These layers synergistically extract meaningful features from the input text, capture linguistic nuances, and generate grading predictions. The layers transform words into low-dimensional vectors, while the convolutional layer captures local features. The LSTM network layer processes sequential data, while the intrinsic feature layer extracts inherent characteristics. The fully connected layer integrates these features to produce final grading scores. The system showed adaptability, scalability, and accuracy, catering to different lengths and complexities of English subjective questions in the test. Through iterative refinement and optimization, the created system offers a robust solution for automating the grading process of English subjective questions, streamlining assessment workflows, and enhancing educational efficiency.