Blank Node Proliferation in RML-Generated RDF: Causes, Consequences, and Controlled Skolemization

Blank node proliferation in RML-generated RDF is a quiet, cumulative failure. It does not throw an error. The output Turtle is syntactically valid. The triplestore accepts the load. Only when developers start writing SPARQL joins across departments, linking graphs from two source systems, or asking why the provenance framework cannot find what it is looking for do the cracks appear — by which point the graph may contain hundreds of thousands of anonymous nodes embedded at structurally critical positions. The five root causes we have identified — surrogate keys, join tables, NULL foreign keys, nested data structures, and processor defaults — are all avoidable with deliberate mapping design. The four-tier IRI minting hierarchy gives practitioners a clear decision tree that works for virtually every relational source pattern they will encounter. The HR case study shows that eliminating blank nodes entirely from a 10,000-record graph requires changing five Subject Maps, produces a 7.5x SPARQL query speedup, and raises cross-source merge success from 21% to 100%. Blank node hygiene deserves the same place in knowledge graph engineering practice that primary key design holds in relational database design or null safety holds in modern programming languages. The techniques for achieving it are well understood, the tooling exists, and the payoff is significant. The main thing that has been missing is a clear statement of the problem and a practical path to fixing it. That is what this paper has attempted to provide.

[1] T. Berners-Lee, J. Hendler, and O. Lassila, "The Semantic Web," Scientific American, vol. 284, no. 5, pp. 28-37, May 2001. [2] W3C, "RDF 1.1 Concepts and Abstract Syntax," W3C Recommendation, Feb. 2014. [Online]. Available: https://www.w3.org/TR/rdf11-concepts/ [3] A. Dimou, M. Vander Sande, P. Colpaert, R. Verborgh, E. Mannens, and R. Van de Walle, "RML: A Generic Language for Integrated RDF Mappings of Heterogeneous Data," in Proc. LDOW at WWW, Seoul, Apr. 2014. [4] A. Hogan, A. Harth, J. Umbrich, S. Kinsella, A. Polleres, and S. Decker, "Searching and Browsing Linked Data with SWSE: The Semantic Web Search Engine," J. Web Semant., vol. 9, no. 4, pp. 365-401, Nov. 2011. [5] M. Arenas, B. C. Grau, E. Kharlamov, S. Marciuska, and D. Zheleznyakov, "Faceted Search over Ontology-Enhanced RDF Data," in Proc. ACM CIKM, Glasgow, Oct. 2011, pp. 1829-1832. [6] P. Heyvaert, B. De Meester, A. Dimou, and R. Verborgh, "Declarative Rules for Linked Data Generation at your Fingertips!" in Proc. ESWC Demos, Heraklion, Jun. 2018. [7] W3C RML Working Group, "RDF Mapping Language (RML) — Draft Community Group Specification," W3C, 2023. [Online]. Available: https://w3id.org/rml/core/spec [8] Morph-KGC Contributors, "Morph-KGC: Scalable Knowledge Graph Materialisation from Heterogeneous Sources," 2023. [Online]. Available: https://morph-kgc.readthedocs.io/ [9] A. Iglesias-Molina et al., "The RML Ontology: A Community-Driven Modular Redesign After a Decade of Experience in Mapping Heterogeneous Data to RDF," in Proc. ISWC, Athens, Nov. 2023. [10] Knowledge Graph Construction Workshop (KGCW), "Proceedings 2022-2024," co-located with The Web Conference. [Online]. Available: https://kg-construct.github.io/workshop/