MULTIPLE PERSPECTIVES ON THE CERRADO: A COMPUTATIONAL ANALYSIS THROUGH TEXT MINING IN ACADEMIC HISTORIOGRAPHICAL PRODUCTION

Abstract

The search for patterns and trends in the historiographical production on the occupation and colonization of the Cerrado in academic narratives, through traditional hermeneutics, constitutes a complex and challenging task, considering the cognitive limitations of humans in accurately analyzing a vast volume of text. On the other hand, the technique called Distant Reading, created by Franco Moretti, proposes distancing oneself from the object of reading, not as an obstacle, but as a form of knowledge, extracting fewer details and more relationships, patterns, and forms within a work or between works. This technique finds operational support in Digital Humanities, an interdisciplinary area that adopts computational methods and techniques to support the development of studies and analyses in the areas of social sciences and humanities. This research proposed to analyze a set of historical documents, organized within a specific framework, to find patterns and trends in the academic narratives reproduced in a textual corpus, using a group of computational tools and techniques through text mining. The results obtained in this study, through systematized procedures in a Proof of Concept, synthesized quantitatively, showed that the use of digital technologies can significantly contribute to the analysis, interpretation, and understanding of historical documents and point to trends in academic narratives. Text mining confirmed the hypotheses raised about the semantic shift from "Sertão" to "Cerrado," with the purpose of discursively reinventing the territory as an economic category and commodity production, and the substantial distancing of narratives between developmental perspectives and the inclusion of Traditional Communities in this process.