Development of AIS Using Simulated Learners, Bayesian Networks and Knowledge Elicitation Methods

The development of adaptive instructional systems (AIS) is an iterative process where both empirical data on human performance and learning, and experimentation using computer simulations can play a role. The paper presents our current efforts to advance adaptive instructional system technology conceived as self-improvement systems [37]. The paper describes our methodological approach for informing the design and implementation of adaptive instructional systems by conducting concurrent research activities using 1) Bayesian networks for modelling learning processes, 2) knowledge elicitation of expert instructors, and 3) simulated learners and tutors to explore AIS system design options. Each activity fulfills separate but complementary objectives. Bayesian networks modelling of learners’ performance provides the means to implement predictions of learners’ performance, and selection of adaptive learning content. Knowledge elicitation methods are fundamental in understanding human capabilities and limitations in the context of AIS systems design that support and regulate the cognitive demands of the learner and instructor. Simulated learner and tutor interactions enable the specification of detailed cognitive process models of learning and instructions.