Análise de visualidades jornalísticas: aplicação de métodos digitais na pesquisa com imagens em jornalismo

Abstract

This work aims to propose methodological approaches to provide researchers in the Journalism field with resources for a broad qualitative visual analysis of journalistic image collections, exploring analysis software, visualization techniques, and possibilites of digital approaches to images. For that, we built a database containing a mapping of the brazilian research about photojournalism in order to identify how this type of image has been analyzed in the country. It led us to recognize that a significant portion of research analyzing journalistic images relies on either methodologies from other, more stablished areas to conduct their research or strictly qualitative approaches for in-depth description of these images, this thesis proposes to apply digital methods which align the quantitative and qualitative aspects of research, considering the characteristics of contemporary journalistic images and their publishing environment. These methods are characterized by studying social phenomena through their insertion into the digital environment, considering its logics and characteristics and adapting to them whenever necessary. The methods applied in this thesis, identified in a systematic literature review and inserted in another database we built on this thesis, focus on images produced in the context of journalistic convergence for publication, circulation, and consumption following the logic of digital platforms, with the entire news production chain being mediated by software. We identified five research stages in works that utilize digital methods for image collection analysis. These stages were replicated in a case study. They start with (1) pre-collection, involving the preparation and design testing of the research; (2) data collection, where the dataset to be analyzed is assembled; (3) dataset refinement, adjusting collected metadata, downloading and resizing images for their (4) analysis, and (5) creating visualizations that will also serve as methodological resources. There are three inputs into the dataset: (1) visualization and analysis based on plastic similarity between images; (2) similarity between content, context, and circulation through the use of Computer Vision APIs; and (3) graphics with images plotted based on metadata collected. After the description, we applied these stages in a case study analyzing how the news vehicle Folha de S.Paulo visually portrayed the 2022 Brazilian presidential elections on its website and Instagram profile. Through this application, we were able to identify potentials, limitations, and possible developments, refinements, and adjustments to the applied methods.