Adaptabilidade temática em sistemas tutores inteligentes híbridos

• Ano de defesa: 2008
• Autor(a) principal: QUINDERÉ, Pedro Sérgio Gomes
• Orientador(a): MARTINS, Weber
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Goiás
• Programa de Pós-Graduação: Mestrado em Engenharia Elétrica e de Computação
• Departamento: Engenharia
• País: BR
• Palavras-chave em Português: Sistemas Tutores Inteligentes Híbridos, regras de especialistas, redes neurais artificiais, adaptabilidade temática
• Palavras-chave em Inglês: Hybrid intelligent tutoring systems, expert rules, artificial neural networks, thematic adaptability
• Área do conhecimento CNPq: CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO::METODOLOGIA E TECNICAS DA COMPUTACAO::ENGENHARIA DE SOFTWARE
• Link de acesso: http://repositorio.bc.ufg.br/tede/handle/tde/976

In the context of efficient information transmission among people and, particularly in the helping of learning and training processes, this investigation presents results on the use of the technology of Hybrid Intelligent Tutoring Systems, based on artificial neural networks and expert rules, developed by Martins [MEA 2004], Melo [MEL 2003] and Meireles [MEI 2003]. Due to the fact that, in its initial empirical validation, neural training data has been originated from courseware in Introduction to Data Processing , some doubts have remained on the applicability of the trained neural network to other scenarios. The present production has approached these issues by the formalization of the content format and by presenting promising empirical results in two other scenarios: Scientific Methodology and Biological Rhythms . Results were analyzed by non-parametric methods with 5% significance. They reinforce the hypotheses that the studied tutoring system is efficient, able to reduce differences of distinct groups and shows thematic adaptability actually