| Resumen |
The impact of the air pollution phenomenon has been long studied, but most often with a fragmented approach, without closely looking at the relationship between different components that characterize it, such as sensor-based data, health data from institutional databases, and data on how it is perceived by human beings in social media. The research developed in this study introduces an integrated methodological framework that analyses sensor data on air pollution distributed in space and time, combined with health data and social data narratives that reflect how different communities perceive this phenomenon in space and time; exploring how these different heterogeneous sources can be combined to better understand the impact of air pollution phenomena at the large-city level in the Valley of Mexico. We introduce a Spatio-temporal data integration and mining framework that aims to discover trends and insights regarding the distribution of the impact of an air pollution phenomenon in terms of human health and perception. The main peculiarity of our methodological framework is the integration of different large data sources by combining a series of methods: NLP (topic modeling), data mining (data cubes, unsupervised learning, and clustering), and GIS capabilities (spatial interpolation, choropleth maps) that together provide a better understanding of the quantitative and qualitative patterns emerging at a different spatial scale and temporal granularity. Overall, this shows how social data, when combined with quantitative data, can provide a better understanding of the impact of a given phenomenon, such as air pollution. © 2022 Roberto Zagal Flores et al. |
