From Beta to RC: The Marriott Reparative Metadata Assessment Tool (MaRMAT)
The Marriott Reparative Metadata Assessment Tool (MaRMAT) is an open-source, Python-based application developed by University of Utah librarians for auditing metadata for potentially harmful terminology. Built with the goal of replacing tedious and bias-prone keyword searching methods, MaRMAT conducts bulk, multifield queries on schema-agnostic, tabular metadata against pre-curated or user-supplied custom lexicons. Producing a report flagging potentially problematic terminology by metadata element, term category, and original context, MaRMAT aids metadata practitioners in the review and remediation process.
Introduced in 2024, MaRMAT Beta was a rudimentary, proof-of-concept application developed in collaboration with AI tools; however, user responses demonstrated a clear need among cultural heritage metadata practitioners for a tool of this kind. Using internal seed funding, a student programmer was hired to refactor the application, increase operating system usability, enhance design and built-in features based on accessibility standards and user feedback, and implement multithreading for increased performance speeds. This talk will focus on MaRMAT-RC’s codebase and core functionality. Detailing the transition from AI-assisted beta programming to a fully human-coded application, we will highlight key challenges encountered during development, outline the team’s decision-making strategies for improving program architecture, and future directions for the project.