Digital Literacy and Knowledge Management in Libraries: AI-Driven Insights into Information Retrieval, User Behavior, and Mental Health

Abstract

Abstract Libraries have transformed into advanced learning spaces where digital literacy and knowledge management are central to enhancing user engagement and overall well-being. Conventional library analytics, however, are limited in their capacity to interpret complex patterns of user behaviour and emotional cues that affect both information-seeking processes and mental health outcomes. This study aims to develop an artificial intelligence (AI)-based predictive framework for academic libraries, designed to strengthen digital literacy initiatives, optimise knowledge management, and provide detailed insights into user behaviour and psychological states. User interaction data were sourced from digital library platforms, including search query logs, access frequencies, and feedback submissions. Incomplete and redundant entries were removed, and textual data were converted into numerical representations using the Term Frequency-Inverse Document Frequency (TF-IDF) method to facilitate computational analysis. The Pearson Correlation Coefficient (PCC) was employed to examine associations among information retrieval patterns, cognitive load features, and emotional responses. The Bitterling Fish Optimizer tuned Decision Forest Tree (BFO-DFT) model, a hybrid machine learning (ML) approach, integrates Bitterling Fish Optimization (BFO) for feature selection with Decision Tree (DT) and Random Forest (RF) algorithms for classification, enabling predictions of user satisfaction and potential stress levels. Findings indicate that the BFO-DFT model demonstrates robust predictive performance and generalisation capacity. Specifically, the model achieved superior metrics with an accuracy of 0.83, recall of 0.87, and F1-score of 0.85 in forecasting user behaviour and satisfaction. Overall, the results highlight that AI-driven optimisation can enhance information retrieval efficiency, foster digital literacy development, and facilitate the early identification of mental stress indicators, thereby supporting intelligent and sustainable library environments.