The European Commission is collecting structured data from many sources, in particular from its member states. There are scenarios where this data is distributed, heterogeneous and published with different schemas. We present how such data is aggregated and integrated at the European Commission using a Knowledge Graph which is called "EU Knowledge Graph". We present the knowledge graph architecture we use, how the data in the knowledge graph is accessed by non-expert users, how it can be accessed using natural language and which services are constructed on top. The approach that we follow is novel with respect to two aspects: first we reuse Wikibase (the software that runs Wikidata) as an infrastructure for the Knowledge Graph and second we make it accessible via natural language using QAnswer. This allows to perform analytic queries by non-expert users.