The CK25 Corporate Knowledge Reference Dataset for Benchmarking Text 2 SPARQL Question Answering Approaches

<p>Question answering over Linked Data / RDF-based Knowledge Graphs is a subdiscipline of Natural Language Processing. The specific challenge of Text 2 SPARQL focuses on translating natural language questions into SPARQL queries, a structured query language for interacting with RDF-based Knowledge Graphs. In this paper, we present a dataset consisting of a knowledge graph, an ontology, and a query catalog, which can be used as a reference for Text 2 SPARQL techniques such as LLM RAG-based approaches. The dataset was created for the First International TEXT2SPARQL Challenge, which we organized as part of the 4th International Workshop on LLM-Integrated Knowledge Graph Generation from Text (Text2KG) at the ESWC2025.</p>