매일 아침, 어제의 AI를 한 통으로 정리해 보내드립니다메일로 받아보기

METAL LAB

Automated item evaluation: Predicting item acceptance and rejection using LLM-generated critiques

arXiv:2608.066092026-08-10

arXiv:2608.06609v1 Announce Type: new Abstract: Automated item evaluation (AIE) refers to the use of computational methods to assess item quality without requiring manual expert review or field testing of the items under evaluation. We aimed to build a near-comprehensive AIE model by predicting item acceptance and rejection from item text using historical rejection data from a large-scale standardized testing program. The dataset contained 52,759 English language arts (ELA) and mathematics items

저자 · Hotaka Maeda, Yikai Lu

arXiv에서 원문 보기