Biases in knowledge representation: an analysis of the feminine domain in Brazilian indexing languages

Sullen Oliveira Milani


The process of knowledge representation, as well as its tools and resulting products are not neutral but permeated by moral values. This scenario gives rise to problems of biases in representation, such as gender issues, dichotomy categorizations and lack of cultural warrant and hospitality. References on women’s issues are still scarce in the literature, which makes it necessary to analyze to what extent the terms related to these particular issues are inserted in the tools in a biased
way. This study aimed to verify the presence of the terms female, femininity, feminism, feminist, maternal, motherly, woman/women within the following Brazilian indexing languages: Subject Terminology of the National Library (STNL), University of Sao Paulo Subject Headings (USPSH), Brazilian Senate Subject Headings (BSSH) and Law Decimal Classification (LDC). Each term identified in the first three alphabetical languages generated a registration card containing both its descriptors and non-descriptors, as well as scope notes, USE/UF, RT, and BT/NT relationships. As for the analysis of LDC, the registration card was filled out by following the categories proposed by Olson (1998). The results showed signs of biases, which enabled the proposition of guidelines that
may contribute to minimize them and open the way for further discussions.

Full Text: