An adaptive semantic retrieval framework for digital libraries integrating graph neural networks, ontology, and user behavior

This paper presents a novel approach to knowledge organization and information retrieval in digital libraries through an adaptive semantic retrieval framework that integrates graph neural networks (GNNs), ontological knowledge structures, and user behavior analysis. Traditional knowledge organization systems often employ static classification methods that inadequately represent multidimensional relationships and fail to adapt to evolving user needs. Our framework addresses these limitations through a unified computational approach that combines formal semantic representations with empirical usage patterns. The system architecture includes ontology-driven knowledge graph construction, multi-relational GNN-based representation learning, comprehensive user behavior modeling, and an adaptive retrieval mechanism that dynamically balances domain semantics with personalized relevance signals. Experimental evaluation across diverse digital library collections demonstrates significant performance improvements, with the integrated framework achieving 81% precision and 85% recall, substantially outperforming conventional retrieval models. The proposed approach enables more intelligent and responsive information discovery while maintaining semantic coherence, offering a promising direction for adaptive knowledge organization that bridges traditional boundaries between formal classification approaches and user-centered design principles.