semantic_subject_classification

To ensure accurate classification:

1. Keyword/Thematic Set: Define a robust mapping dictionary including key terminology, common academic sub-fields, and potential core authors for each subject.
2. Standardized Mapping: Use a strict lookup dictionary where keys are subject categories and values are the exact folder names:
   • LLM -> LLM
   • Trapped Ion -> trapped_ion_and_qc
   • Black Hole -> black_hole
   • DNA -> DNA
   • Default -> music_history
3. Logic Flow: Perform semantic evaluation for the first 4 subjects. If the similarity score falls below a defined threshold (e.g., 0.3) for all 4, assign the file to music_history as the absolute fallback.