Aligning Author-Supplied Keywords for ETDs with Domain-Specific Controlled Vocabularies

Abstract

Subject access can provide essential points of access for users to find, identify, select, and obtain various resources available in libraries. Subject access is not always available, however, due to the increasing amount of metadata created by non-catalogers (including author-supplied metadata), changes in libraries’ discovery services, and a lack of best practices for aligning non-controlled vocabularies to authorized subject headings. This paper addresses the issue of author-supplied metadata, specifically how to align keywords submitted by authors of electronic theses and dissertations (ETDs) with Library of Congress Subject Headings (LCSH) and discipline-specific taxonomies by analyzing 32,696 keywords from 5,365 master's theses and doctoral dissertations submitted to the University of Illinois at Urbana-Champaign's institutional repository between 2010 and 2014. This paper shares findings from the data analysis, including that matching rates vary depending on college, with newer or rapidly-developing fields (such as the School of Molecular and Cellular Biology) having lower matching rates than traditional, well-established fields of study (such as the College of Agriculture, Consumer, and Environmental Sciences), and recommends that when keyword reconciliation is performed, it should be done with more than one authority in tandem for the best results; when the LCSH and discipline-specific controlled vocabularies were combined, matching results were slightly or moderately increased.