Measuring Intersectional Biases in Historical Documents
Authors:
Nadav Borenstein,
Karolina Stańczak,
Thea Rolskov,
Natália da Silva Perez,
Natacha Klein Käfer,
Isabelle Augenstein
Abstract:
Data-driven analyses of biases in historical texts can help illuminate the origin and development of biases prevailing in modern society.
However, digitised historical documents pose a challenge for NLP practitioners as these corpora suffer from errors introduced by optical character recognition (OCR) and are written in an archaic language. In this paper, we investigate the continuities and tran…
▽ More
Data-driven analyses of biases in historical texts can help illuminate the origin and development of biases prevailing in modern society.
However, digitised historical documents pose a challenge for NLP practitioners as these corpora suffer from errors introduced by optical character recognition (OCR) and are written in an archaic language. In this paper, we investigate the continuities and transformations of bias in historical newspapers published in the Caribbean during the colonial era (18th to 19th centuries). Our analyses are performed along the axes of gender, race, and their intersection. We examine these biases by conducting a temporal study in which we measure the development of lexical associations using distributional semantics models and word embeddings. Further, we evaluate the effectiveness of techniques designed to process OCR-generated data and assess their stability when trained on and applied to the noisy historical newspapers. We find that there is a trade-off between the stability of the word embeddings and their compatibility with the historical dataset. We provide evidence that gender and racial biases are interdependent, and their intersection triggers distinct effects. These findings align with the theory of intersectionality, which stresses that biases affecting people with multiple marginalised identities compound to more than the sum of their constituents.
△ Less
Submitted 21 May, 2023;
originally announced May 2023.
Coalescence, the thermal model and multi-fragmentation: The energy and volume dependence of light nuclei production in heavy ion collisions
Authors:
Paula Hillmann,
Katharina Käfer,
Jan Steinheimer,
Volodymyr Vovchenko,
Marcus Bleicher
Abstract:
We present results of a phase space coalescence approach within the UrQMD transport and -hybrid model for a very wide range of beam energies from SIS to LHC. The coalescence model is able to qualitatively describe the whole range of experimental data with a fixed set of parameters. Some systematic deviations are observed for very low beam energies where the role of feed down from heavier nuclei an…
▽ More
We present results of a phase space coalescence approach within the UrQMD transport and -hybrid model for a very wide range of beam energies from SIS to LHC. The coalescence model is able to qualitatively describe the whole range of experimental data with a fixed set of parameters. Some systematic deviations are observed for very low beam energies where the role of feed down from heavier nuclei and multi-fragmentation becomes relevant. The coalescence results are mostly very close to the thermal model fits. However, both the coalescence approach as well as thermal fits are struggling to simultaneously describe the triton multiplicities measured with the STAR and ALICE experiment. The double ratio of $tp/d^2$, in the coalescence approach, is found to be essentially energy and centrality independent for collisions of heavy nuclei at beam energies of $\mathrm{E_{lab}}> 10 A$ GeV. On the other hand the clear scaling of the $d/p^2$ and $t/p^3$ ratios with the systems volume is broken for peripheral collisions, where a canonical treatment and finite size effects become more important.
△ Less
Submitted 16 March, 2022; v1 submitted 13 September, 2021;
originally announced September 2021.
