Synthesizing Didactic Explanatory Texts in Intelligent Tutoring Systems Based on the Information in Cognitive Maps

The paper deals with the problem of synthesizing texts that explain decisions of an Intelligent Tutoring System for the learner. Solving this problem is important for increasing human trust in the ITS decisions by finding effective arguments and synthesizing explanatory texts. We describe an approach to preparing and concentrating raw data that is similar to the process used for mapping. The texts explaining the system's decisions are generated using the method of parametric maps, which are visualized as Cognitive Maps of Knowledge Diagnosis: the transition from particular maps to the combined map and the development of a system of arguments based on these data. We show the structure of an explanatory text and the accompanying visualization of a cognitive map. We demonstrate the explanation synthesis on the example of a graduate student in the course "Simulation modeling". The explanation consists of the text and supporting visualization. Preliminary analysis has shown significant student interest in receiving explanations, containing verbal and non-verbal (cognitive maps) components, from the Intelligent Tutoring System about its decisions.