User Search Terms and Controlled Subject Vocabularies in an Institutional Repository

Abstract

Controlled vocabularies are an important mechanism for ensuring consistency in a repository and necessary for maximal collocation for searching by subject. Of the metadata fields used for retrieval, subjects are particularly valuable, allowing for a type of collocation less easily achieved through titles or abstracts. However, the quality of subject terms can vary based on policies guiding their selection. If controlled vocabularies present a solution for reducing metadata 'noise', one must also consider the search behavior of the user. How well do user queries align with a controlled vocabulary, and what's the level of effort required to reconcile legacy subject terms with a new vocabulary? Our analyses uses search queries which led users to items in the KU ScholarWorks institutional repository. These queries are reconciled against FAST (Faceted Application of Subject Terminology) terms, legacy subject terms, and more broadly across repository records to determine the potential effects on search behavior. The effort required to reconcile legacy metadata will be considered as the repository seeks to reconcile its history of uncontrolled language with a more systematic and extensible vision for the future. Presented at Open Repositories 2015, Indianapolis, IN, June 10, 2105.