Abstract
The rapid expansion of biomedical literature presents significant challenges for efficient information retrieval in health sciences libraries and digital repositories. This study presents a user-centered experiment evaluating the retrieval efficacy of controlled vocabularies (specifically Medical Subject Headings, or MeSH) compared to user-generated tagging (folksonomies) and a integrated hybrid model within biomedical knowledge organization systems. Thirty-six participants, comprising biomedical researchers, clinical practitioners, and graduate health sciences students, completed twenty standardized retrieval tasks across three experimental system configurations. System performance was assessed using both objective system metrics (Precision@10, Mean Average Precision, Recall, and Task Completion Time) and subjective user experience evaluations (System Usability Scale and NASA-TLX cognitive workload assessment). Results demonstrate that while controlled vocabularies yield significantly higher retrieval precision (p < 0.001) for precise clinical queries, user tagging demonstrates superior recall and lower task latency when navigating emerging biomedical topics and non-standard clinical terminology. The hybrid system, which combined formal ontology structure with dynamic social tags, achieved the highest overall F1-score (0.842) and superior user satisfaction ratings. These findings indicate that modern biomedical knowledge organization systems should move beyond binary choices, implementing hybrid architectures that leverage the structural rigor of controlled vocabularies alongside the adaptive flexibility of user tagging.
