Reparative Metadata Assessment with MaRMAT: A New Tool for Identifying Harmful and Outdated Language

Identifying potentially harmful or outdated language in existing metadata is the first step toward reparative metadata practices. However, useful, open, and accessible metadata assessment tools are few and far between. The Marriott Reparative Metadata Assessment Tool (MaRMAT) was developed by librarians in collaboration with ChatGPT to streamline the identification process so that more time and attention can be dedicated to the remediation process. The purpose of this session is to introduce users to MaRMAT: how it was developed and how it can assist metadata professionals in the bulk analysis of collections metadata to identify problematic terms in controlled and uncontrolled fields to promote remediation toward current and preferred terminology. MaRMAT conducts a bulk, multifield query on schema-agnostic, tabular metadata against a pre-curated or user-supplied lexicon to produce a report flagging potentially problematic terminology by metadata element, term category, and original context to aid in the review and remediation process. Therefore, MaRMAT replaces tedious individual keyword searching methods that are often prone to selection bias. This session will introduce attendees to MaRMAT and its key features, providing an opportunity to test the program using sample metadata and curated lexicons for outdated Library of Congress Subject Headings and biased language. This tool is openly available on GitHub and runs in a GUI on Windows or in Terminal on MacOS; the session will provide training in both versions and discuss a variety of potential use cases.