Context-aware Japanese furigana using Sudachi and ModernBERT

This project introduces a context-aware furigana (reading annotation) system for Japanese text, leveraging Sudachi for morphological analysis and ModernBERT for disambiguation. It is designed for Japanese language learners, NLP researchers, and developers of reading tools who need accurate, context-dependent readings for homographs. The approach is interesting because it moves beyond simple dictionary lookups to achieve state-of-the-art accuracy in resolving ambiguous kanji readings, directly improving readability and comprehension in digital Japanese content.