Dataset Catalog

GeoQuestions1089 is a crowdsourced geospatial question‑answering dataset containing 1,089 triples of natural‑language questions, SPARQL/GeoSPARQL queries, and answers, targeting the YAGO2geo knowledge graph. The dataset is split into two parts: GeoQuestions_c (1,017 entries without linguistic errors) and GeoQuestions_w (72 entries with grammar, syntax, or spelling errors). Version 1.1 introduced several improvements, including unified query format, corrected natural‑language case handling, query classification fixes, and replacement of erroneous triples. Questions are categorized into nine groups covering various aspects of geospatial QA.