Principle

Apply the divide & conquer stragegy more generally to focus on what actually makes a difference no matter the number of pages.

Example of important pages

Note that those too should ideally also apply divide & conquer.

  1. this one?
  2. MostImportantEquations
  3. LinkOfTheMonth and LinkOfTheYear
  4. Cosmology#CurrentModel
  5. Beliefs#List
  6. ImprovingPIM#TrackMovingGoals
  7. ? ExtendedLayeredModel
  8. ? StructuralInformationAsymmetries

Constraints

  • remain dynamical since the change of one page importance has consequences for its related page
  • automated ranking might not be sufficient as some behavior might not reflect the importance of the page
    • e.g. knowing that a topic is important yet failing to improve or not yet dedicated the time to do so (so frequency of update or size are not enough)
    • allows to distinguish between explicit goals and actual behavior that might or might not participate to it

Solution

boost (or lower) ranking per page

  • via dedicated explicit parameter as PTV
    • fixed range e.g. from -1 to +1
      • but it forces to re-order previous one as a new one get added
    • dynamic and then normalized within the pre-defined range
      • but forces to really know what other scores have been given
    • default value would be neutral
  • via existing related PTVs
    • startrecall, recall, startprepare, ...
  • via graph metrics
    • see existing groupkeypages.php and groupstats.php.

groupkeypages.php has been update to have access to related PTVs but is not yet using them in a composite metrics.

Note that this should also be applied to ImprovingPIM#VisualDecayOfInformation.

ToRefactor