The need for rules to structure data

• Rules for e.g. Gaussian Input and Output need extensive human perception (training)
• Outputs tend to be:
  Context/Keyword/Version dependent/Unexpected extensions/Not self-defining
• Closed software often needed to link Input/Output/Visual models/Editing
  ("rules" are often hard encoded and not enumerated)
• Humans must handle the terminology (synonyms, acronyms, semantics, ontologies)
• Ontologies/dictionaries CAN be organised (Crystallography/CIF)