Logo do repositório

Percorrer por autor "NOYA, MIGUEL FREITAS"

A mostrar 1 - 1 de 1
  • Simulador de Carreiras Inteligente
    Publication . NOYA, MIGUEL FREITAS; Matos, Telmo Manuel Sampaio Pinto
    Choosing the right career is one of the most important decisions within anyone’s life. This moment is also often clouded by doubt, uncertainty and a general lack of adequate guidance. The main goal of this project is to solve this problem, by developing the "Intelligent Career Simulator". This tool is a cutting-edge, data-driven system created to assist users in researching job possibilities and coming to wise decisions. The simulator takes advantage of hundreds of thousands of past student decisions by utilizing data from the EDULOG and Brighter Future Foundations to provide customized recommendations based on each user’s distinct interests and objectives. This dissertation was developed in partnership with Business to Future (B2F). B2F is a company specialized in the field of Business Intelligence (BI) that was challenged by EDULOG to develop an Artificial Intelligence (AI) solution to be integrated into their ecosystem. This tool is designed to assist students who are about to finish their studies, students who want to choose a study area or just someone who wants to change their field of expertise and wishes to transition smoothly into the new market. EDULOG is a think tank of the Belmiro de Azevedo Foundation focused on the field of education, dedicated to the research, analysis, and discussion of the Portuguese education system. The project’s mission is to contribute to the strategic planning of education in Portugal, aiming for excellence in education. To organize this study, a state-of-the-art analysis was conducted to select the best approaches and technologies. Based on this analysis, two scenarios for the ICS were created and tested: in Scenario 1 the AI Bot queries the standard database (developed for other analytical needs), in scenario 2 the bot queries a database optimized specifically for an AI agent. To analyze the results, both qualitative and quantitative measures were used. The results show that using the optimized database allows the bot to be more accurate and provide better results in most of the cases. The general improvements in response quality were calculated to be around 18% on average, and, in terms of financial viability, the optimized solution saves around 0.002$ per query on average, demonstrating a payback period of around five months. This study brings new insights on how to optimize the usage of data for smart AI agents, and their respective limitations. With the ICS, the goal is to empower users, by providing detailed instructions for those pursuing specific careers, offering clear steps to achieve individual goals.
    2026-01-23Dissertação de mestrado Acesso aberto Ver mais