Open Science Grant for Andreea Iana and Martin Böckling
SpatialBenchRAG: A Multimodal, Multilingual Benchmark for KG-grounded RAG over Diverse Geographies
The project SpatialBenchRAG introduces the first multimodal, multilingual benchmark for evaluating Retrieval-Augmented Generation (RAG) systems grounded in spatial Knowledge Graphs (KGs). Addressing the lack of standardized resources for spatial reasoning, the benchmark combines diverse geographic datasets, including OpenStreetMap and Wikidata, to support realistic, context-aware evaluation across tasks such as spatial question answering, multi-hop reasoning, and visual queries.
SpatialBenchRAG will be released as an open-source resource on HuggingFace and Zenodo, featuring multilingual datasets in five languages, standardized evaluation protocols, and supporting evidence for rigorous assessment. By enabling transparent and reproducible research, this initiative aims to accelerate the development of geographically grounded AI systems for applications like disaster response, geo-aware assistants, and spatial fact-checking.
The Open Science Office of the University of Mannheim funds projects that are intended to foster reproducible open science. One of those projects has been awarded to Andreea Iana and Martin Böckling for the SpatialBenchRAG project.

