An intelligent tutoring system authoring tool, checker, and platform development that reproduces the classroom lecture styles of human teacher

This paper describes a method for developing intelligent tutoring systems (ITSs) that reproduce the classroom teaching format as taught by human instructors. We have already published monographs on the intelligent tutoring system framework and platform development. In those monographs, we propose a hierarchical graph partitioning method in a subject to facilitate graph construction and prevent inconsistencies in graph construction, which is represented as an extension of the knowledge graph. However, there were areas that were not developed in depth in the implementation using Hidden Markov Models (HMMs) and Markov Decision Process (MDP). This paper (book) will delve deeply into the problems raised in developing intelligent tutoring systems that replicate the classroom teaching format as taught by human instructors. What it means to reproduce the classroom teaching format as taught by human instructors is not about gathering multiple students in a classroom for uniform instruction, but rather replicating the teaching format exactly. This includes reproducing elements such as the display of picture data on TV, the associated video playback, the interpretation using human teacher’s voice about the picture data displayed on TV, the question and response. Our intelligent tutoring system consists of an intelligent tutoring content authoring tool, a checker (or player) for the authored intelligent tutoring content, and a platform. Therefore, we propose the development of an authoring tools, a checker (or player) development, and a platform development method. We also propose a pedagogical module development approach for three tutoring models, such as the one-to-one tutoring model, the tutoring model using the knowledge graph, and the tutoring model using Partially Observable Markov Decision Process (POMDP). In this paper, we propose an acyclic violation checking algorithm that can occur if we construct a hierarchical partition-type knowledge graph using a graph drawing tool and a way to hierarchically partition the knowledge graph. Automatic construction of knowledge graphs requires a vast amount of educational digital resources. We also propose a method to manually construct hierarchically partitioned knowledge graphs using visualized graph drawing tools included in knowledge graph authoring tools, if there is no or lack of extensive pedagogical digital data.