Empowering Users Through Democratic Indexing: Evaluating Folksonomy and Taxonomy search terms for Improved Information Retrieval in Koha ILS

Abstract

The study explores the use of democratic indexing and its effectiveness at providing controlled access to information inside Koha Integrated Library Systems (ILS). By contrasting traditional approaches of using controlled vocabularies and taxonomies with 'developer folksonomies', we show how the inclusion of natural language tagging increases subject resource discoverability without compromising the controlled vocabulary structure. Our analysis of 30 social science pairings (with 190 unique tags overall) demonstrates a 13% difference in recall citing 85% vs 72% for hybrid indexing, while the precision difference cited 78% vs 88% respectively was of minimal importance because of ambiguous synonymy. Our findings reveal key points such as: 1) folksonomies have tangible value in disciplines in the humanities to provide more expressive discovery (ex. 91% total recall in Sociology), 2) users prefer to use natural language tags as evidenced by 82% of users being satisfied when citing natural language tags and, 3) moderated hybrid models are the best approach for providing a degree of flexibility - however not too much, as accuracy becomes paramount. The study introduces a weighted ranking algorithm (70% taxonomy/30% folksonomy when applying it to STEM disciplines) and strategies for implementation and customization of Koha using subject auto-tagging creativity and speficic moderation of cataloged material. Our overall efforts contribute to the goal of advancing participatory librarianship through empirical evidence in hybrid indexing as a preferred alternative method that can provide equitable access without collapsing to bibliographical stiffness.