Question

If a system uses Bayesian Knowledge Tracing (BKT) to model student learning, what specific probability calculation is updated each time a student answers a new problem?

Accepted Answer

Each time a student answers a new problem, a Bayesian Knowledge Tracing system updates the probability that the student has mastered the specific skill being tested. This value is known as the probability of mastery. The calculation uses Bayes&#x27; Theorem to refine this estimate based on the student&#x27;s most recent performance. If the student answers correctly, the system updates the probability of mastery upward, accounting for the chance that the student guessed correctly despite not knowing the material. If the student answers incorrectly, the system updates the probability of mastery downward, accounting for the chance that the student made a slip despite actually knowing the material. To perform this update, the system relies on four fixed parameters: the prior probability of mastery, which is the student&#x27;s estimated knowledge before any problems; the transition probability, which is the chance a student learns the skill after attempting a problem; the guess probability, which is the chance of getting a correct answer without mastery; and the slip probability, which is the chance of getting an incorrect answer despite having mastery. By combining these parameters with the student&#x27;s new observation, the system computes the posterior probability of mastery, which then becomes the starting probability for the next problem.

Home → All Courses → Education Courses → Learning Technology Engineering → Flashcard

If a system uses Bayesian Knowledge Tracing (BKT) to model student learning, what specific probability calculation is updated each time a student answers a new problem?