Algorithmic Discriminations and Their Ethical Impacts on Knowledge Organization: A Thematic Domain-Analysis

Abstract

Personal data play a fundamental role in contemporary socioeconomic dynamics, with one of its primary aspects being the potential to facilitate discriminatory situations. This situation impacts the knowledge organization field especially because it considers personal data as elements (facets) to categorize persons under an economic and sometimes discriminatory perspective. The research corpus was collected at Scopus and Web of Science until the end of 2021, under the terms “data discrimination”, “algorithmic bias”, “algorithmic discrimination” and “fair algorithms”. The obtained results allowed to infer that the analyzed knowledge domain predominantly incorporates personal data, whether in its behavioral dimension or in the scope of the so-called sensitive data. These data are susceptible to the action of algorithms of different orders, such as relevance, filtering, predictive, social ranking, content recommendation and random classification. Such algorithms can have discriminatory biases in their programming related to gender, sexual orientation, race, nationality, religion, age, social class, socioeconomic profile, physical appearance, and political positioning.