GLP definition. We have modelled the flow of analytics on a proof-of-concept implementation to better understand the current GLP model. Because of this implementation, we propose that every node in the GLP include a JSON structure similar to the one below, defining the limits (including thresholds) and the contributions of this node. { // from the GLP: limits; if absent, full-success is implied limits: {
GLP definition. This section attempts to provide a complete specification based on our current understanding, and factoring in feedback from a previous version of this document. As the GLP is a tree structure-based game definition, the activities created in order to analyse what happens along the gameplay will require some information to be propagated between child and parent nodes: we refer to this as propagation / knowledge update. The specification spreadsheet does not describe when the propagation occurs; this is addressed by defining a run-time model which describes the exact steps to take in response to user actions. UCM has actually built a working run-time proof-of-concept to test different hypothesis on how the GLP will work, and document our findings here. Regardless of GLP implementation choices, it is important to make the analysis and the game sequencing mirror each other, such that, for example, if GLP conditions on a node make the analysis mark it as failed, the same conditions must trigger the restart of the node inside the game to avoid incoherent results. Analytics is not sequencing, and cannot control what nodes are visited when. Therefore, even though analytics may detect that certain nodes have been "failed" and cannot be completed successfully (for example, due to excessive attempts; or because a sub-node needs to be repeated), analytics cannot force nodes to be avoided or repeated. The only choice for analytics is whether to keep or discard incoming data. Proposed changes and clarifications to the GLP spreadsheet From feedback for the previous version of this document, we understand that: ● In the Score Distribution sheet, ○ Number of Failures => Number of Attempts ○ Good Answers Total => Score ● Activities return scores and maximum-possible scores when completed. The scores are used to contribute, using teacher-defined weights found in the GLP (see Score Distribution, Math Example) towards the activities’ Objectives (either Learning Objectives or Competencies), and the contributions are then termed Activity Points. So a score of 10 in an activity that contributes 30% towards LO1 and 70% towards LO2 would yield 3 Activity Points in LO1 and 7 in LO2. This calculation to obtain Activity Points from score and competencies percentages is done from the analysis side. ● Scores for activities (and per-activity activity maximum scores, needed to calculate percentages of maximum) can be propagated to ancestors, allowing their success to be calculated along...
