Pentágono da privacidade no big data privacy analytics: proposta de modelo multifacetado de garantia da privacidade e do valor analítico

Abstract

The continuous increase in the amount of data generated and available has created an environment characterized by a gigantic mass of data, which grows not only in terms of volume and quantity, but also in terms of variety, being created and moving at high speed. If before, the focus and availability of data had as a priority interface the so-called structured data, today they are in much smaller quantity and importance, and adjustments and improvements in technologies and analyzes were partly carried out to adapt to this new reality which is conventionally called big data. The value created by big data analysis has brought many positive factors to different fields, but, on the other hand, it has also brought some problems. One of the issues of great relevance in this context that has raised great concern in society are the threats to privacy, brought about by advanced analyzes of large volumes of data. The question raised as a result of several researches is that the procedures, techniques and technologies currently available cannot fully guarantee privacy. As well as legislation that regulates data protection and privacy, such as in the Brazilian case the General Data Protection Law (LGPD). On the other hand, it is possible to obtain a very high level of privacy, but at the cost of nullifying the possibilities of extracting value from big data. Given this complex scenario, the focus of this research is to propose a multifaceted model (because it needs to be composed of multiple approaches and constructs) within the scope of big data analytics, which guarantees privacy, at the same time, in which it does not make it impossible to extract value. The methodology proposed for this work will be based on a quantitative approach, using text mining and unsupervised machine learning with a focus on creating clusters. After the quantitative phase, the privacy policies of 28 companies were analyzed using a qualitative approach, with two main objectives: finding the best number of clusters; and obtain an understanding of their distinctive characteristics, in order to improve the analysis carried out. The privacy policies of the thousand largest companies in Brazil were selected, classified as such by the Valor1000 ranking (2021). The results pointed to the consistency of the Privacy Pentagon model in the proposed Big Data Analytics; as well as what was called “opaque privacy” in this work.