AI RESEARCH
Improving Lexical Difficulty Prediction with Context-Aligned Contrastive Learning and Ridge Ensembling
arXiv CS.AI
•
ArXi:2605.08950v1 Announce Type: cross Lexical difficulty prediction is a fundamental problem in language learning and readability assessment, requiring models to estimate word difficulty across different first-language (L1) backgrounds. However, existing approaches rely on regression-only