A Novel Framework using Large Language Models to Automate Coursework Feedback for Computer Science modules

Prompt and sufficient feedback is essential for students' academic learning since it enables them to review their learning techniques and improve their areas of weakness. Nevertheless, delivering personalised feedback to every student continues to be difficult for teachers due to its demanding and time-intensive nature. While automated feedback systems are available, their primary focus is providing feedback on a single subject, and most of them utilise statistical analysis or traditional machine learning techniques to provide feedback. Moreover, no feedback model utilises the same criteria to generate text-based feedback for more than one subject. Generative artificial intelligence (GEN AI) has recently made incredible progress, and large language models (LLMs) can retain the context from the vast amount of text. Hence, this research presents a framework that employs an innovative technique to offer text-based feedback to students in different fields of study. This framework employs two LLMs, one for generating the feedback and another for categorising it into separate subjects using suitable headings for structural organising. Consequently, the output produced by this technology corresponds to the original tone of the teacher.