Skip to content

Latest commit

 

History

History
86 lines (59 loc) · 1.84 KB

README.md

File metadata and controls

86 lines (59 loc) · 1.84 KB
Raw

Transition Matrix based Simulators

Env Protocol

  • What are provided as environment parameters

    • action space
    • knowledge structure

  • What is the action space, knowledge or item?

    • knowledge
    • item

Additional

There are two different mode for

  • no_measurement_error
  • with_measurement_error

Learner

  • Which type is the learner, infinity or finite?
    • infinity
    • finite
  • Which mode is the response of the learner, real or trait?
    • real
    • trait

Item

  • What are the types of items in this environments?

    • learning item
    • test item

  • Is the learning item base same with the test item base?

    • Yes
    • No

  • What are included in an item?

    • content
    • knowledge
      • single
      • multiple
    • attribute
      • guessing
      • slipping

  • What is the relation between item and knowledge in learning item?

    • one-to-one
    • one-to-many
    • many-to-one
    • many-to-many

  • What is the relation between item and knowledge in test item?

    • one-to-one
    • one-to-many
    • many-to-one
    • many-to-many

Reward

  • Step reward: $R(t) = \sum_{k=1}^{K}[\alpha_k(t+1)-\alpha_k(t)]$
  • Episode reward: $G=\sum_{t=0}^{T-1}R(t)$

Find these two equations in [1] (Eq.(1) and Eq.(2))

Agent Protocol

  • What is recommended (i.e., what is the action space), knowledge or item?
    • knowledge
    • item

Systems

  • binary: Study I in Tang et al. [1]
  • tree: Study II in Tang et al. [1]

Original Code

Tang et al. [1] did not provide source code.

Reference

[1] Tang X, Chen Y, Li X, et al. A reinforcement learning approach to personalized learning recommendation systems[J]. British Journal of Mathematical and Statistical Psychology, 2019, 72(1): 108-135.