Modelo logístico para quantificação da latência brega na música brasileira

• Ano de defesa: 2024
• Autor(a) principal: Oliveira, Jefferson da Silva
• Orientador(a): Andrade, José Ailton
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Link de acesso: http://repositorio.ufc.br/handle/riufc/80015
• topic: Teoria de resposta ao item; Modelos bayesianos; música brega (música); Discurso Lítero-musical; Item response theory; Bayesian models; Brega music (music); Literary-musical discourse

The cheesy music is present in Brazilian mainstream music, especially in the 70s and 80s. However, even though it’s often treated as something minor and with little poetic influence, it’s notable that cheesy music has influenced other musical movements. The initial objective of this study is to apply Item Response Theory to quantify the cheesy latency in Brazilian music, identifying historical context characteristics in its musical and textual planes. Considering these aspects, a questionnaire will be developed based on Item Response Theory, considering dichotomous responses with the aim of measuring the latent cheesy trait. To do this, we will use a Bayesian model based on IRT for parameter estimation, assigning a probability distribution to the items and identifying in the corpus which cheesy songs from the selected 70s and 80s have markedly cheesy characteristics. All analyses will be based on estimates obtained via MCMC and through computational packages. The results obtained throughout the study show that the songs related to cheesy music are understood through their discursive description in their textual plane, meaning that the measured cheesy latent traits correspond to the textual characteristics described in the songs. On the other hand, it was difficult to identify these characteristics through the musical plane, in the identification of sound instruments in the songs under study.