Text Mining Bibliographic Metadata for Inclusivity: Analyzing Most Frequent Words in Titles, Summaries, and Subjects

Abstract

Academic libraries have embraced diversity, equity, and inclusion (DEI) principles as core tenets for serving their users. Many of these libraries have undertaken a diversity audit of their collections, evaluating content as well as authorship and amending acquisition processes to increase representation of historically marginalized groups. Techniques used in an audit can include comparison to bibliographies and peer institutions, but few libraries have used text mining of bibliographic metadata to uncover the inclusivity of their collections. This article describes one such study, performed at Raritan Valley Community College, to determine whether language displayed in the title, summary, and subject fields was inclusive and welcoming to library users. Prompted by a new functionality available for WorldCat Discovery that would allow for local updates to problematic subject headings, the process involved uploading MARC metadata to Voyant Tools to learn the most frequent terms in each bibliographic field. Results demonstrated that while the metadata includes welcoming language, improvements could be made by updating subject headings, deaccessioning outdated titles, and educating users in navigating the library catalog.