If done with sensitivity, autonoding need not be a "foul practice".

A good autonoder would work like this:

  • User marks up source material into nodes, giving each node a sensible name (e.g. "Luke 7:15" (qv)). Some hardlinks may be hand-inserted, but many may be ommitted.
  • Script compares each word and each phrase of (say) three words or less, with its own list of nodenames, and if there is a match, converts the phrase into a hardlink.
  • Script searches E2 for a node matching every word and every phrase (below a given length) in the document, and inserts hardlinks where appropriate. Stopwords would be required, and possibly some human intervention.
  • A new login name is created for the autonoder, e.g. "Roger's Profanisaurus", so that:
    • Human does not take credit for autonoded content
    • Autonoded content is easily identifiable
    • Admin may delete autonoded content with a single SQL command if everything goes tits-up.
  • Nodes are created.
Any old fool can export junk into E2: it takes effort to export quality nodes. (Effort which so far I have failed to stump up).