ABSTRACT
This chapter covers the challenges encountered while working with a bilingual dataset of news articles in English and Portuguese published in the United Kingdom and Brazil. It presents the limitations of data availability in Brazil as well as the lack of natural language processing (NLP) tools with reasonable performance in Portuguese language. It focuses mainly on two NLP techniques: Sentiment Analysis and Named Entity Recognition. The media coverage of London 2012 and Rio 2016 Olympic legacies is used as case study. The chapter’s main contribution refers to the design of a methodology for critical use of data and software, using explainable AI tools to interrogate the algorithms’ outputs.
