bikiDATA: A Python Library to Query and Explore Large-Scale RDF Datasets

Abstract

While knowledge graphs offer unparalleled data flexibility, the semantic gap between RDF triples and the native objects used by software engineers remains a significant barrier to entry. Developing knowledge-graph-backed applications typically requires deep expertise in SPARQL and complex data-mapping layers. To lower this threshold, we present bikiDATA: a high-performance storage solution and a Python library engineered for the modern software developer. Unlike traditional wrappers, bikiDATA abstracts the complexities of the RDF data model into a developer-friendly API that feels native to the Python ecosystem. Beyond standard SPARQL support, the system provides a comprehensive suite for production-grade applications, including integrated full-text search, knowledge graph embeddings, and visual similarity search. Already in use in ongoing projects at FIZ Karlsruhe, bikiDATA reduces integration complexity, improves scalability, and enhances query performance. The source code and executable demo notebook are publicly available at https://github.com/ISE-FIZKarlsruhe/bikidata.

Demo paper accepted at 23rd European Semantic Web Conference (ESWC) May 10-14 2026 Dubrovnik, Croatia