Abstract
Introduction. The rapid growth of scientific literature calls for robust methods of quantitative analysis and domain mapping. Existing science mapping workflows typically project the original bipartite “document–term” relation into a single-mode network of pairwise co-occurrence links, visualizing a similarity proxy rather than the original metadata structure. This limits the analysis of structural relations explicitly encoded in the metadata. Objective. This study demonstrates the methodological potential of Formal Concept Analysis (FCA) as an intermediate analytical layer between bibliometric metadata and their visual representations, and evaluates visual models that preserve the bipartite structure of the data. Materials and methods. A corpus of publications on Data Visualization was retrieved from the IEEE Xplore database (IEEE Terms field, 2022–2026; 5,456 records, 46,722 term occurrences). IEEE Terms were selected as an expert-verified controlled vocabulary, eliminating the need for term normalization. Concept lattices were constructed using In-Close4 and In-Close5, with cross-verification of the results. Visualization was performed using Graphviz, Scimago Graphica, VOSviewer, and Sankey diagrams. Results. A lattice of 5,740 formal concepts was constructed. Different slices of the same lattice support complementary analytical representations: the concept lattice foregrounds hierarchical inclusion relations, the co-occurrence network shows pairwise term associations, and the Sankey diagram preserves three-part “publication–concept–term” relations. A stable thematic core was identified linking Data visualization and Visualization to Real-time systems, Internet of Things, and Image sensors, alongside peripheral clusters such as Protocols, HTTP, Communication systems, Filtering, Circuits, and Feedback. The compatibility of FCA-derived data with widely used bibliometric tools was also established. Conclusion. FCA can serve not only to generate formal concept lattices but also to select structured and interpretable subsets of bibliometric data for subsequent visualization. The proposed pipeline provides traceability from derived concepts to underlying records and complements existing network-based approaches.