This paper presents a comparative analysis of the data and metadata models used by national libraries that publish their collections following the principles of Linked Open Data (LOD). Using the Crosswalk method, the study mapped the classes and properties adopted by different institutions, identifying conceptual convergences, terminological variations, and distinct modeling strategies. The results reveal heterogeneous practices, ranging from simplified models to robust, ontology-based structures. Despite shared core entities such as Work and Person, semantic differences remain regarding the scope and level of detail. The findings support the development of semantic alignment strategies and shared vocabularies to improve interoperability.
