Enhancing Automated Assignment Assessment in Higher Education with a CNN–BiLSTM Model

Educational technology theories have lagged behind the rapid development of relevant technologies. Due to advancements in information and communication technologies, college students can now reach audiences beyond their professors. Assignments are now being graded based on podcasts, infographics, blog posts, and social media marketing. In addition to improving their theoretical and practical abilities, students are able to reach a wider audience with the use of these instruments. Improved assignment evaluation was achieved by a multi-stage prediction model utilising CNN and BILSTM. Latent Dirichlet Allocation was used for feature extraction after data was prepped by removing duplicate and incomplete information. To examine school records, the CNN-BILSTM model applies geographic information gleaned from traffic records. With a Recall of 95.56% and an F-Measure of 99.49%, the model performed admirably while evaluating assignments. A more precise assessment of assignments may be possible with the use of sophisticated DL techniques, according to these results. The necessity for ongoing technological advancement in educational assessment is emphasised by the fact that ICT-based communication tools enhance students' engagement with real-world audiences.