Text enters the model
The system identifies sentence boundaries and converts English writing into language units the model can analyse while preserving paragraph and contextual relationships.
Word/PDF scans first exclude contents pages, references, tables and non-prose, then send only qualifying English prose to the detector.
