Sentiment analysis and topic modeling of Portuguese and Brazilian song lyrics through the years

• Autor(a) principal: D´Alva, Inês Mariana da Trindade
• Data de Publicação: 2023
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10071/30770
• topic: Music lyrics; Análise de sentimentos -- Sentiment analysis; Topic model; Text mining; Letras de música; Modelação de tópicos

Music lyrics are a rich source of information, within the various components in the musical context. With their distinctive identity and narrative elements, lyrics have the power to convey profound messages, with the emotions and sentiments they portray and the themes addressed. Over time, these lyrical components have evolved, mirroring the changing dynamics of society. This dissertation aims to study these sentiment and topic changes in the national scope of Portugal and Brazil, spanning from the 1960s to the 2020s. To achieve this, we employ a lexicon-based approach for sentiment analysis and utilize BERTopic and LDA for topic modeling. The results of our research reveal an emotional contrast between the two countries. Brazilian songs predominantly exude positivity and uplifting sentiments, while Portuguese songs often carry a prevailing undertone of negativity. The extracted topics from the lyrics frequently align with each nation’s historical and societal experiences. However, some instances show a disconnect, where lyrics do not accurately mirror challenging periods in terms of topics or sentiment polarities. This suggests that lyricists may employ their musical creations as a form of escape from reality.