Namespace LMKit.Nlp

ScriptFamily: Coarse Unicode script classification used by multilingual text-quality heuristics. Every supported written language maps to exactly one entry, so comparing the ScriptFamily of two characters tells whether a token mixes scripts (a strong OCR-noise fingerprint regardless of language), and lets script-specific heuristics gate themselves: word-morphology signals are only meaningful for alphabetic scripts, while CJK and abjad scripts rely on script consistency and distribution checks (see DominantScript).