国际教育成就评价协会（IEA）：一种基于人工智能的TIMSS项目验证与对齐方法（英文版）

Abstract This study investigates the feasibility of using artificial intelligence (AI) to automate the alignment of assessment items with the Trends in International Mathematics and Science Study (TIMSS) framework. Specifically, it evaluates prompt-based large language models (LLMs) and supervised machine learning approaches across three classification dimensions: content domain, cognitive domain, and item difficulty. To support this analysis, a comprehensive, structured item bank was constru