SCORE DISTRIBUTION MONITORING IN AI-BASED JAPANESE SPEAKING ASSESSMENT: AN EXPLORATORY DATA-ENGINEERING PERSPECTIVE
Article Information
Abstract
Artificial intelligence (AI) supports scalable speaking assessment, but
reliable operation also requires monitoring the structure of generated data. Sparse
item coverage and concentrations in specific score categories can complicate
subsequent analysis. This study examines category balance and response
coverage in AI-scored Japanese speaking data, focusing on operational dataquality indicators rather than item-response parameter estimation.
Keywords
no

