🎯 Enhanced Knowledge Extraction Pipeline
Major improvements to how we extract curriculum content from uploaded documents. The new pipeline better filters out irrelevant metadata (headers, footers, page numbers) and focuses on actual educational content.
- Curriculum relevance scoring (0.0-1.0) now filters non-educational content
- "Exam test" principle: only extracts content students would be tested on
- Improved handling of lecture slides with speaker notes
- Fixed: Flashcards no longer include document metadata or file paths