---

6.5 Tag normalization ¶

TagNormalizer maps incoming tag strings to a canonical vocabulary. Reasons:

• Lowercase + hyphenated consistently ("Family" ⇒ "family"; "home office" ⇒ "home-office").
• Synonyms collapse to a canonical form ("fam", "families" ⇒ "family").
• Garbage gets dropped — empty strings, characters outside the allowed alphabet, single-character tags.

The map lives in src/main/precompiled/ai/ownsona/TagNormalizer.java as plain static data. To add a synonym, edit the map and rebuild. No migration needed — existing rows keep their original tags, and the normalizer applies only to new writes.