Filtragem baseada em comentários para recomendação de recursos educacionais em plataformas de conteúdos diversificados

• Ano de defesa: 2024
• Autor(a) principal: Carvalho, Henrique Carlos Fonte Boa
• 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: Aprendizado de Máquina; Machine Learning; Comentários; Comments; Objetos de Aprendizagem; Learning Objects; Recursos Educacionais; Educational Resources; Sistema de Recomendação; Recommender System; Youtube; CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO; Computação; ODS::ODS 4. Educação de qualidade - Assegurar a educação inclusiva, e equitativa e de qualidade, e promover oportunidades de aprendizagem ao longo da vida para todos.
• Link de acesso: https://repositorio.ufu.br/handle/123456789/43592; http://doi.org/10.14393/ufu.te.2024.654

The technological evolution has promoted an increasingly connected society, facilitating interaction among people and the massive sharing of content. This advancement positively impacts various areas of knowledge, including education, where the amount of available materials grows exponentially. However, this abundance of educational resources brings challenges, such as the difficulty in identifying and choosing the most suitable ones amidst a vast array of content. These challenges are even greater in non-strictly educati- onal repositories, such as Wikipedia, LinkedIn, YouTube, TikTok, Vimeo, among others, where content is shared by users from different areas and interests, including educational materials. This work innovates by developing an approach that uses user comments along with AM techniques to recommend Learning Objects (LO) in environments with diverse content. For this, the most frequent vocabularies in each class, educational or non-educational, were used. Two variations were developed: the rigid variation and the flexible variation. The rigid variation uses Machine Learning (ML) algorithms to classify videos as educational or non-educational based on the most frequent vocabularies, recommending videos that the algorithm is certain” are educational. The flexible variation classifies each comment individually as educational or non-educational, analyzing the classification of all comments on the videos and recommending them with a certain degree of certainty” of belonging to the educational class. The results obtained revealed that comments are, in fact, an excellent feature for the classification of LOs, especially when using the most frequent vocabularies of each class. Experiments indicate that the approach allows identifying LOs with an impressive accuracy of 95%. Additionally, the flexible variation demonstrates greater adaptability to work with the real world, enabling better recommendation of materials with different quantities of comments. Finally, the LOIS was developed, a Recommendation System (RS) that assists teachers and students in Virtual Learning Environment (VLE) in finding LOs on YouTube.