Sistema web de simulados com correção automática, análise de desempenho e recomendações inteligentes baseadas em IA

Abstract

This document aims to present the development of a web system featuring a list of simulated entrance exams, including automatic correction, performance analysis, and personalized study suggestions, using artificial intelligence to perform searches and generate automatic recommendations. The proposal arose from the need to offer an accessible educational tool that encourages active and adaptive learning through immediate feedback and continuous support. The distinguishing feature of the project is the artificial intelligence module, which uses the Ollama API with the Phi-3 model to generate personalized study recommendations based on users' incorrect answers, returning a list of suggested topics and content for review. The methodology adopted was evolutionary prototyping, allowing the system to be built through continuous cycles of testing, evaluation, and improvement. With each version of the prototype, new adjustments were incorporated to enhance the interface, the user flow, and the overall student experience, ensuring a more natural development process centered on experimentation. Because it is focused on personalized study recommendations and automated corrections, the project was named SCAR-I (Simulados, Correção automática, Análise de desempenho, Recomendações inteligentes, IA), which in a certain way resembles a "scanner."