At the“19th International Semantic Web Conference”(ISWC), Timo Homburg won the Best Student Paper Award. We extend our warmest congratulations to him!
Timo Homburg, together with Prof. Dr. Steffen Staab of the University of Stuttgart—who is supervising his dissertation—and Dr. Daniel Janke of the University of Koblenz-Landau, submitted a paper titled “GeoSPARQL+: Syntax, Semantics, and System for Integrated Querying of Graph, Raster, and Vector Data.”
Abstract model of a raster with surrounding geometry, various cells and their values (the pixels), their surrounding geometries, and a legend for the semantic interpretation of the pixel values; graphic: Timo Homburg, CC BY-SA 4.0
In it, the authors present an approach to the semantic representation and querying of raster data in a Semantic Web graph. They extend the GeoSPARQL vocabulary and query language to support raster data as a new type of geodata. They extend the query language with new filter functions, raster algebra functions, and functions that can compare vector and raster data. They illustrate their approach using use cases with real-world datasets. Finally, they describe a prototype implementation and validate the feasibility of their approach.
Ontology model for representing a raster as an extension of the existing GeoSPARQL ontology, graphic: Timo Homburg, CC BY-SA 4.0
One example is the city of Cologne, which was flooded by the Rhine. The streets are represented by vector data, while flood levels are represented by raster data. In a real-world use case, a query would, for example, search for all street segments that are not covered by more than 10 cm of water. This is only possible if the data model can represent raster data, vector data, and semantics (street, water, depth, 10 cm) and allows for a joint query of these representations.
Existing geographic information systems lack the ability to represent semantics. The GeoSPARQL standard [6] currently does not describe any representation of raster data or any query options for it. Systems that currently support geographic information on the Semantic Web cannot process raster data and therefore do not allow such queries.
The results of this work are intended to be incorporated into the next version of the GeoSPARQL standard. Timo Homburg has already submitted a corresponding request to the Open Geospatial Consortium.
