Natural language generation for intelligent tutoring systems

Abstract

This project tackles the problem of automatically generating text for a conversational intelligent tutoring system (ITS). The conversational ITS gives questions to the student and then analyzes their answer using machine learning algorithms. When the student’s answer is classified as being incorrect, the system must give an appropriate hint to help them answer the question or understand the problem. The system is powered by a database of questions, answers, and hints. The component generating the hints is called the “hint generation” model. To this end, we propose a feedback and hint generation model capable of generating meaningful, personalized, pedagogical hints. To make the course more engaging and effective for each student, we propose a “content generation” model capable of generating new content, including questions, answers, and hints, automatically from unstructured text (such as Wikipedia) or interactively with assistance from teachers. In addition, we also investigate how the system may be capable of efficiently adapting its content and tutoring strategies to individual students.

Date	16 May 2022
Original language	American English
Awarding Institution	École de technologie supérieure
Supervisor	Sylvie Ratté (Supervisor)