Toward Educational Explainable Recommender System: Explanation Generation based on Bayesian Knowledge Tracing Parameters

Abstract

Explainable recommendations, which provides explanations about why an item is recommended, helps to improve the transparency, persuasiveness, and trustworthiness. However, there are few research in educational technology that utilize explainable recommendation. Previous research has identified that learner motivation can deteriorate while using educational recommender systems, and that providing additional forms of feedback can improve performance and increase trust. In this paper, we propose an explanation generator using the following parameters from Bayesian knowledge tracing models: guess (giving a correct answer despite not knowing the skill) and slip (knowing a skill, but giving a wrong answer) for a quiz recommended by the system. Recommended quizzes were categorized into different feature types according to the value of the model parameter and explanation texts are generated based on these feature types.