Projeto e desenvolvimento de um algoritmo de recomendação aplicado ao sistema science

Nowadays, a big part from the softwares that provides some service, relies on a recommender system. A recommender system can be defined as an algorithm that generates a recommendation, based, weather on the ratings from the users, or based on the content from the service. The algorithms are differen...

Descripción completa

Detalles Bibliográficos
Autor: Oliveira, Victor Benoiston Jales de
Tipo de recurso: tesis de grado
Estado:Versión publicada
Fecha de publicación:2021
País:Brasil
Institución:Universidade Federal Rural do Semi-Árido (UFERSA)
Repositorio:Repositório Digital da Universidade Federal Rural do Semi-Árido (RDU)
Idioma:portugués
OAI Identifier:oai:repositorio.ufersa.edu.br:prefix/7096
Acceso en línea:https://repositorio.ufersa.edu.br/handle/prefix/7096
Access Level:acceso abierto
Palabra clave:Sistema de recomendação
Filtragem colaborativa
Filtragem baseada no conteúdo
Aprendizado de máquina
CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA E TECNOLOGIA
Descripción
Sumario:Nowadays, a big part from the softwares that provides some service, relies on a recommender system. A recommender system can be defined as an algorithm that generates a recommendation, based, weather on the ratings from the users, or based on the content from the service. The algorithms are different on its filtering model. There are three kinds of data filtering, collaborative filtering, content-based filtering, and hybrid filtering, we are going to use the content base filtering on this recommender system development for improving the SCIENCE. The problem taken in this article, was the kNN (k-Nearest Neighbours), unsupervised machine learning algorithm. The metric used, was the cosine similarity, for that reason, the closest similarity possible, based on the angle of similarity between two vectors. This given project, presents the development and integration of recommender system using content-based filtering, and the development of future algorithms that uses collaborative filtering.