ľ¹Ï¸£ÀûÓ°ÊÓ

On Thursday 10 July, will defend his PhD dissertation 'Words Matter: A Computational Toolkit for Charged Terms'. In his dissertation, Brate creates a computational toolkit for examining charged terms associated with colonialism.

Decolonisation in museum metadata

Museums have not historically been neutral spaces. In the West, many major museums were founded during the colonial era, and their collections and institutional histories are often deeply intertwined with imperial expansion. Today, however, many such institutions are working toward decolonisation strategies and initiatives.

These initiatives include reviewing metadata to spot biased narratives, identifying socially biased or harmful terms, and explaining such biases to readers. These initiatives are typically carried out on an ad hoc basis by working groups. While extremely valuable, such approaches do not easily scale up to the massive digital datasets that these institutions have.

Toolkit of methods

In this dissertation, Brate aims to support the work of decolonisation by using data-driven insights derived from large text sources. His goal is to reveal the social bias of potentially culturally insensitive terms.

Brate proposes scalable approaches, such as statistical, machine learning-based, and generative AI-driven approaches. These are targeted at specific social signals in texts. This toolkit of methods, which uses publicly available datasets from literary fiction, news, and social media, corresponds well with known negative social biases associated with harmful terms in the museum domain.