PsyScore: A Psychometrically-Aware Framework for Trait-Adaptive Essay Scoring and ZPD-Scaffolded Feedback
Researchers have developed PsyScore, a novel framework designed to enhance automated essay scoring and provide adaptive instructional feedback. This system integrates psychometric principles with neural network architectures, specifically incorporating the Graded Partial Credit Model (GPCM) into a neural scorer. PsyScore also features a feedback generator that tailors its strategies based on a student's diagnosed ability level, aiming to improve pedagogical alignment. Experiments on the ASAP++ dataset indicate that PsyScore performs competitively in scoring while offering more relevant feedback. AI
IMPACT This framework could improve educational tools by providing more accurate and personalized feedback to students.