Intelligent Virtual Assistant for students based on personalized scheduling and diagnostic assessment

Abstract

TutorIA, an intelligent virtual assistant, is being developed to improve the academic performance of university students through personalized study recommendations. This initiative arises from the difficulties many students face in organizing their time, selecting appropriate materials, and adapting to increasing academic demands. The system is implemented using a microservices architecture, where a backend service is designed to manage user and intelligent agent data, while these services are consumed by a mobile application developed in Flutter. The agent integrates a reinforcement learning model using the Deep Q-Learning algorithm, which progressively adapts study guides based on student performance. This performance is evaluated using tests proposed by the agent itself, allowing for the identification of areas of difficulty and the adjustment of future recommendations in a personalized manner. The result of this project has largely met its initial objectives, including: the generation of a diagnostic assessment test, the development of a machine capable of learning from student errors, the implementation of mechanisms to explain subject-specific content, and the creation of a personalized study schedule tailored to the user’s availability and needs.