In GeoQA we have created the richest (in terms of geospatial knowledge) knowledge graph, YAGO2geo, by extending the well-known knowledge graph YAGO2 with data found in multiple geospatial datasets (geographical administrative data provided by a range of official sources, the Global Administrative Areas dataset and OpenStreetMap). We, also, develop scalable techniques and software for keeping YAGO2geo up-to-date. The GeoQA project makes this valuable knowledge accessible to both experts and non-experts, by developing a prototype question answering engine, based on natural language processing and knowledge graph embedding techniques. To enhance accessibility to non-experts, the engine will present to the end-users intuitive visualizations of the answers. This will be the first such engine internationally to be able to answer effectively and efficiently geospatial questions.

Geographic knowledge has been studied for many years by researchers in Geography, Geographic Information Systems (GIS), Databases, Artificial Intelligence and the Semantic Web, and there is a wealth of research results concerning representation, querying and inference for geographic knowledge. However, the current large knowledge graphs contain very little knowledge about the spatial attributes (e.g. geometries) of the geographic features (e.g. Greece). At the same time, very popular search engines, like Google, are struggling to give immediate answers to complex geographic questions like “Which cities are within 100 km of Berlin”. There is only one question answering system in this domain, which has been developed by this group and forms the starting point of this project.