Uma abordagem híbrida apoiada por algoritmo bioinspirado e tecnologias de web semântica para recomendação personalizada de objetos de aprendizagem

• Ano de defesa: 2021
• Autor(a) principal: Pereira Júnior, Cleon Xavier
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Uberlândia; Brasil; Programa de Pós-graduação em Ciência da Computação
• Palavras-chave em Português: Sistemas de Recomendação; Recommendation Systems; Objeto de Aprendizagem; Learning Object; Filtragem Baseada em Conteúdo; Content Based Filtering; Inteligência Artificial; Artificial Intelligence; CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO; Computação; Inteligência artificial - Aplicações educacionais; Web semântica
• Link de acesso: https://repositorio.ufu.br/handle/123456789/32725; https://doi.org/10.14393/ufu.te.2021.463

The Adaptive and Intelligent Educational Systems (AIES) area is constantly evolving and is working to apply recent technologies to create personalized learning environments. For AIES development, Artificial Intelligence (AI) techniques are widely explored and commonly combined with pedagogical theories. This work aims to contribute to the area of AI applied to education, by presenting an approach that uses Semantic Web technologies and a bio-inspired algorithm to perform personalized recommendation of Learning Objects (LO), using Content-Based Filtering (CBF). In this research, we combine Virtual Learning Environment (VLE) repositories and materials available on the Web (Youtube and Wikipedia) to cover topics of a given content with materials in different formats. Regarding the materials on the Web, these are retrieved and structured as LO. We implemented the approach in the Classroom eXperience (CX) VLE and also created an extension for Moodle. Experiments were conducted with this implementation. One experiment compared three bio-inspired algorithms with two different databases and, after analysis, concluded that the genetic algorithm performs satisfactorily. Other experiments aimed to analyze the students' opinions regarding the recommendation. Students positively evaluated the recommendation that took into account their level of knowledge and offered additional material from a given content. Another experiment considered three different recommendation processes to observe preference possibilities. The recommendations took into consideration the use and non-use of learning styles in the process. The overall average rating was relatively better disregarding the use of learning styles, but there was no statistical significance.