Development and Application of Web-based Machine Learning Program for Automated Assessment Model Generation

Abstract

In order to enable customized learning using artificial intelligence, a scoring model that can evaluate students' responses is needed. Human-scored data and programming skills training artificial intelligence are required to generate scoring models. When teachers develop scoring models, it is very useful to secure many scoring models that can be used in schools. However, the biggest challenge for teachers to create scoring models is programming skills training artificial intelligence. Thus, in this study, we developed a web-based automatic assessment model generation program that can rapidly generate assessment models using graded descriptive responses using supervised learning, without programming. Web program development was created by dividing an analyzer, classifier, web development, and server for extracting features. By developing an assessment model with actual scored data using the developed program, it was determined that an assessment model is reliable. Using the developed program, teachers can create their assessments using scored data without prior programming experience. In addition, the developed program can be used to strengthen teachers' artificial intelligence capabilities.