{
 "operator": "normalise",
 "note": "Official data: the two repair sources, both created by another transformation (gen_coregraph.py), plus the standard repair rules.",
 "entries": [
  {
   "kind": "derived",
   "name": "the dictionary",
   "source": "../../data/world.json#tokens",
   "why": "the universe's own forms — ed1 repairs only into words the document actually says"
  },
  {
   "kind": "derived",
   "name": "the thesaurus",
   "source": "../../data/world.json#stems",
   "why": "the stem families computed from the document; a stranded form is pulled to its family's most-said member"
  },
  {
   "kind": "standard",
   "name": "repair rules",
   "data": {
    "min_length_for_ed1": 4,
    "order": [
     "dictionary (ed1)",
     "thesaurus (stem family)"
    ],
    "tie_break": "alphabetical"
   },
   "why": "deterministic repair; dictionary before thesaurus"
  }
 ]
}