AI-Driven Decision Support in Public Administration: An Analytical Framework

The use of artificial intelligence has been intensifying in public administration, which demands smart tools for several types of analyses. These analyses range from using texts from internal and external documents, for instance, scraped from the web; data from internal structured sources, such as relational databases; and multimedia-originated data, for instance, considering videos, images, and sound recordings. This work aims to present a framework based on the combined analysis of structured and unstructured (textual) data, as well as using geographic information. Two cases are presented: (i) corpora dedicated to public security in Brazil, with some examples of dedicated analyses, and (ii) a spatial-temporal analysis of public opinion distribution about the spread of misinformation about the COVID-19 vaccination campaign in Brazil. Some possible advances in the framework are also commented on, showing how new types of media can be applied in addition to those already used.