Predicting Emotional Impact on Peer Review, Peer Assessment, and Self Assessments Using Deep Learning and NLP in STEM Education
DOI:
https://doi.org/10.58459/icce.2024.4976Abstract
Digitalization has transformed academic environments with an increased volume of scientific articles, projects, reports, and workplace events overload. Applications like NLP, ML, DL, AI, and Cloud computing have ushered in more tools for researchers to automate scientific tasks which has led to a skyrocket in duty overlord and pressure on peer review. Due to huge task overloads, emotional expressions on academic assessments, have outweighed factual justifications consistency. This study highlights the present and future changes associated with academic assessment work overload and how this impacts future scientific lifecycles. Aim: The study investigated the impact of the increasing pressure and burden on academic assessment overload, academic assessment feedback, and increased emotional sentiments on peer reviews. The study applied DL and NLP to investigate the impact of emotional sentiments on academic assessments. A total of 90 reviews and feedback from conferences, journals, and patent comments were utilized to predict 4 years' future sentiment level in academic evaluation. Based on the selected reviewers' comments, our model with an accuracy of 72% predicts a fall in reviewers' emotional sentiments throughout the next four years. The study concluded that Peer Review, Peer Assessment, and Self-Assessment as integral parts of AI-driven sustainable education, are fast integrating with technological evolution to better serve the interest of scientific communities.