C H O I C E S

You can encode (mark up) almost anything.
But what should you encode?

And to what depth?

Your choices will be dictated by ...

• the nature of the material
• the character of your incoming data
• the amount of your funding
• the patience of your funders
• (how much time left till your retirement)
• the scale of the project (how many items)
• the scope and variety of the project
• the purpose of the project
  • desired functionality
  • expected (or guessed-at) audience
  • potential for repurposing
  • potential for sharing/reuse
• your own knowledge or ignorance
• the existence of standards
  • why to avoid them. They are:
    • complex, hard to use
    • not tailored to material
    • not supported by local expertise or compatible with local systems
    • not as good as what you can come up with yourself
    • etc.?
  • why to use them. You can
    • leverage community expertise
    • share data
    • share tools
    • entertain at least a faint hope of "sustainability"
  • library practice summarized in GUIDELINES
    • a good starting-point
    • provide good suggestions rooted in actual practice
    • do not merely define but apply tags, with examples
    • offer five 'levels' of commitment:
      1. raw OCR marked off into pages, linked to page images
      2. = LEVEL 1 + chapter divisions and headings
      3. = LEVEL 2 + refinements. Text may (?) stand on its own.
      4. Better text (keyed or corrected), tagged enough to stand alone
      5. = LEVEL 4 + considerable manual intervention based on subject knowledge.

  Our own projects as examples:

  • (I) Mass-digitization projects Making of America and the Google-scanned books going into HathiTrust
  • (IV) The Text Creation Partnership
  • (V) Middle English Dictionary (and Compendium)
  • (IV) Knight's American Mechanical Dictionary
  • (V) The Faculty CV project

  These differ in

  • their adherence to standards
  • their labor-intensity
  • their longevity (the price of success?)
  • their scale and scope

  But share a common rationale:

  intelligible display intelligent navigation contextually useful search restrictions constraint by method and cost susceptability to incremental improvement