Previous: Cover & differentiate Up: A task independent description of role limiting methods Next: Extrapolate from a similar case

Propose & revise

Here we discuss the original Propose and Revise (P&R) RLM used in the SALT shell [Marcus, 1988]. P&R is traditionally applied to configuration tasks, but can also be used for other types of tasks, like planning, scheduling, assessment, diagnosis. Also the task control is not necessarily task-type specific.

For the different tasks a renaming of the domain ontology has been provided with modelling operators.

When comparing C&D and P&R it appears that different domain ontologies, domain schemata, tasks and inferences are used in P&R and C&D; C&D uses pre-stored solution classes (explanations) to classify the problem instance. P&R also uses pre-stored problem-solution associations, but gradually constructs a solution by using concrete partial and interacting solution elements. These are pairs containing a single requirement or constraints associated with a possible solution: a value for a dimension of the `design'. The result is that domain organisation, tasks and inference relations are all different.

___________________________________________________

• Domain knowledge
  • Domain ontology
  • Domain ontology requirements
  • Case models
  • Domain schemata
• Task Knowledge
• Inference knowledge
• Modelling operators
  • P&R applied to configuration
  • P&R applied to scheduling
    • P&R assumptions rephrased for scheduling
  • P&R applied to diagnosis
    • P&R assumptions rephrased for diagnosis