In the last decade, a large market for location-based services and geospatial applications has emerged. Multiple cartographic providers propose their visualization tool for displaying points of interests. However, the data describing spatial entities is often incomplete and contradictory from one provider to another, thus limiting further applications such as geospatial data mining. Recent works in entity matching tackle this issue by discovering correspondences between spatial entities that refer to the same real world location. To evaluate and compare these works, we propose GeoBench, a tool which facilitates the building of a benchmark for spatial entity matching.