gift-syntax/dsd-tutor-schema

for GIFT quizes

Based on del Soldato, T. & du Boulay, B. (1995) "Implementation of Motivational Tactics in Tutoring Systems," Journal of Artificial Intelligence in Education, 6(4): 337-78:

Problem:
  description: text
  initial_steps: SET < answer, step > # enabled by default at problem onset
  similarity: LIST < problem, integer > # degree of similarity
    # intransitivities can be used for broader/narrower distinctions
  difficulty: LIST < problem, success_predictor_tried, success_pred_untried >
    # regression coefficients can be guessed apriori or derived from practice
  INSTANCE success: untried or [0..1] # partial credit
  assessments: SET < minimum_success, text >
    # the greatest assessment not exceeding success(problem) shown when needed
Answer:
  check: pattern # such as a regular expression describing an answer
    # In general, if more than one answer checks, perhaps more input is 
    # needed to disambiguate.  If none is available, then the system 
    # might just ask for specification.
  meaning: [ accomplishment | giveup | help | agree | disagree ]
    # I use "agree" and "disagree" instead of "positive" and "negative" here
  confidence: integer
    # in speech, confidence can be derived from prosody or relative loudness
Step:
  repetition_limit: posint
  enables: SET < answer, step > # this step is a prereq of these steps
  disables: SET step # this step precludes these steps
    # steps remain enabled unless they are explicitly disabled, I suppose
  success_level: [0..1] # assigned to success(problem)
  present_step_help: help
  next_step_help: LIST < step, help > # only enabled steps are helped with
  surprise_result: help
  INSTANCE repetitions: nonnegint
  INSTANCE previous: step # this could be a list for tracking cycles
  feedback: comment or content_type # custom or stock feedback
    # make sure this fits into the rules as designed
Help:
  general: text or macro script which may refer to problems or steps
    # macros allow referring to another problem only if it has been completed
  specific: text or macro script which may refer to problems or steps
    # specific help ought to be able to call a (simple) sub-problem
  INSTANCE acceptance: integer # the degree to which help was accepted
Comment:
  [ content-type: text ]  # may be omitted where they can be taken from stock.
    # Text here ought to allow variable substitution if not full-fledged macros.
    # Comments should also be added to provide reports to human instructors.
# more work needed for surprise_result, text/macro (HTML?), and set specification