A carregar...
Publicação
Qualidade do ar em zonas urbanas: recolha de dados e o seu reflexo no comportamento online
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 2.37 MB | Adobe PDF |
Orientador(es)
Resumo(s)
Esta dissertação tem como objetivo analisar a relação entre qualidade do ar e os sentimentos expressos nas redes sociais, com foco no episódio de poeiras do deserto do Saara que afetou Lisboa em março de 2022. Para isto, foram recolhidos dados ambientais da plataforma QualAR, e dados meteorológicos do Meteomanz, complementados com a recolha de publicações do Twitter (plataforma atualmente conhecida como X) geolocalizadas em Lisboa. Os textos das publicações foram traduzidos e analisados com o algoritmo Valence Aware Dictionary and sEntiment Reasoner (VADER), permitindo identificar padrões emocionais e avaliar possíveis correlações com a variação dos poluentes atmosféricos. Foi também realizada uma análise preditiva no estudo, usando algoritmos de Machine Learning, nomeadamente Decision Tree e Random Forest, de forma a explorar relações entre os diferentes parâmetros e apoiar a interpretação dos resultados. Os resultados mostram que, apesar de o evento ter provocado níveis excecionalmente elevados de partículas inaláveis com diâmetro < 10 μm (PM10), não se verificou uma alteração significativa nos sentimentos expressos online, que permaneceram maioritariamente neutros ou ligeiramente positivos. Estes resultados sugerem que a perceção ambiental da população não se reflete de forma clara nas redes sociais, possivelmente devido a fatores culturais, sociais ou ao reduzido impacto do Twitter no contexto português. Conclui-se que a integração entre dados ambientais e sociais é metodologicamente viável e relevante, mas requer mais investigação, com períodos mais longos, diferentes cidades e técnicas de análise mais avançadas. Este estudo contribuiu para a compreensão da relação entre poluição atmosférica e perceções públicas em Portugal, colmatando uma lacuna existente na literatura.
Abstract: This dissertation aims to analyze the relationship between air quality and the sentiments expressed on social media, focusing on the Saharan dust episode that affected Lisbon in March 2022. To achieve this, environmental data were collected from the QualAR platform, and meteorological data were obtained from Meteomanz, complemented by the collection of geolocated Twitter posts (now known as X) in the Lisbon area. The content of these posts was translated and analyzed using the Valence Aware Dictionary and sEntiment Reasoner (VADER) algorithm, enabling the identification of emotional patterns and the assessment of possible correlations with fluctuations in atmospheric pollutants. A predictive analysis was also conducted using Machine Learning algorithms, specifically Decision Tree and Random Forest, to explore relationships between the various parameters and support the interpretation of the results. The findings indicate that, although the event caused exceptionally high levels of inhalable particulate matter with a diameter < 10 μm (PM10), there was no significant change in the sentiments expressed online, which remained mostly neutral or slightly positive. These results suggest that the public's environmental perception in not clearly reflected on social media, possibly due to cultural or social factors, or the relatively limited influence of Twitter in the Portuguese context. It is concluded that the integration of environmental and social data is both methodologically feasible and relevant, but further research is needed, involving longer time periods, different cities, and more advanced analytical techniques. This study contributes to the understanding of the relationship between air pollution and public perception in Portugal, addressing a gap in the existing literature.
Abstract: This dissertation aims to analyze the relationship between air quality and the sentiments expressed on social media, focusing on the Saharan dust episode that affected Lisbon in March 2022. To achieve this, environmental data were collected from the QualAR platform, and meteorological data were obtained from Meteomanz, complemented by the collection of geolocated Twitter posts (now known as X) in the Lisbon area. The content of these posts was translated and analyzed using the Valence Aware Dictionary and sEntiment Reasoner (VADER) algorithm, enabling the identification of emotional patterns and the assessment of possible correlations with fluctuations in atmospheric pollutants. A predictive analysis was also conducted using Machine Learning algorithms, specifically Decision Tree and Random Forest, to explore relationships between the various parameters and support the interpretation of the results. The findings indicate that, although the event caused exceptionally high levels of inhalable particulate matter with a diameter < 10 μm (PM10), there was no significant change in the sentiments expressed online, which remained mostly neutral or slightly positive. These results suggest that the public's environmental perception in not clearly reflected on social media, possibly due to cultural or social factors, or the relatively limited influence of Twitter in the Portuguese context. It is concluded that the integration of environmental and social data is both methodologically feasible and relevant, but further research is needed, involving longer time periods, different cities, and more advanced analytical techniques. This study contributes to the understanding of the relationship between air pollution and public perception in Portugal, addressing a gap in the existing literature.
Descrição
Dissertação apresentada à Escola Superior de Tecnologia do Instituto Politécnico de Castelo Branco para cumprimento dos requisitos necessários à obtenção do grau de Mestre em Engenharia Informática - Área de Especialização em Desenvolvimento de Software e Sistemas Interativos.
Palavras-chave
Qualidade do ar Redes sociais Análise de sentimentos Twitter Poeiras do Saara Air quality Social media Sentiment analysis Sahara dust