Can natural language help to understand the origins and persistence of inequality in our society?
The domain of inequality is heavily characterized by causal claims. Multiple researchers in economics, social sciences and humanities have studied the sources of modern inequality.
However, modelling their automatic detection in natural language is still an open issue. This micro-project aims to tackle this gap by detecting explicit causal relationships in Italian texts related to inequality. The output is a system that performs close reading analysis of documents that have been preprocessed by state-of-the-art NLP tools and use Frame Semantics as meaning representation.
● Italian Corpus about inequality extracted and annotated.
● Open source Italian Causation Frame Extractor using a computational construction grammar approach
This micro-project aims to test and further expand the methodology by Beuls, Van Eecke, and Cangalovic (2021) applied for English causations. Moreover, it proposes an improvement to the cited methodology by implementing additional frames to account for the frames of negation and adding information about the tense and aspect implied by the Causation lexical unit. The information on causality given by the system could be beneficial to formulate and verify hypotheses about the drivers of socio-economic inequality.
Van Trijp R., Galletti M., Blin I. (Sony CSL),
Morbiato A., Santagiustina C. R. M. A., Steels L. (VIU),
Beuls K., Van Eecke P. (VUB).