Título: Rede Neural Artificial em Sistemas Tutores Inteligentes Híbridos Baseados em Tipologia Psicológica: Implementação e Análise Empírica
Autores: Melo, Francisco Ramos de; Martins, Weber; Meireles, Viviane; Nalini, Lauro Eugênio G.
Resumo: This paper presents an Intelligent Tutoring System based on Multilayer Perceptron neural networks. It is adaptive and reactive and offers customized and dynamic teaching. Features of user psychological profile are employed as basic elements of customization, and they are complemented by (human) expert rules. The investigation follows the development process of the intelligent tutor and analyses the collected data with respect to the neural network role in the proposed system. Descriptive analysis of data indicates that the application of proposed techniques is adequate, based on empirical results. The main aspects that have been studied are retention (“learning improvement”) normalized gain, navigation total user time and number of steps (length of visited content). The Pearson’s Correlation indicates that the use of neural network is determinant on customization of learning steps (navigation on course contents).
Código DOI: 10.21528/CBRN2005-100
Artigo em PDF: CBRN2005_100.pdf
Arquivo BibTex: CBRN2005_100.bib