Skip to content

Knowledge Graph Discovery

Timo edited this page Jan 21, 2023 · 5 revisions

The SPARQLing Unicorn plugin allows the discovery and exploration of a given knowledge graph resource. This page introduces the ways to do this.

We use the AtlantGIS triple store data (https://github.com/kacebe/AtlantGIS/blob/master/screenshots/atlantgis_sites01.png) as an example in this entry.

Discovery using the Geoconcepts Tab

The Geoconcepts tab shows the result of classes that are known to have at least one instance with a geometry representation attached.

image

Double-clicking any entry in the GeoConcepts tab will load a layer of all geometry instances associated with geometry into QGIS. If a coordinate reference system is defined in the knowledge graph, it will be set as the coordinate reference system of the layer.

image

The context menu of every entry in the GeoConcepts Tab allows for collecting more information about the instances before querying the layer to QGIS. In particular, it allows the viewing of common attributes associated with a concept. These attributes do not have to be present at EVERY instance and may be present multiple times at one instance.

image

The above image shows the "data schema view" of one GeoConcept. DatatypeProperties which are connected to an immediate value, are shown in green. Blue Properties show relations to other instances in the knowledge graph. More specialized properties, e.g., for Geometries, labels, or descriptions, are also highlighted. For each property, a short preview of the values allows an assessment of the data before querying it in the same window.

image

The related concepts dialog allows viewing the connections of the selected GeoConcept to other concepts in the knowledge graph. In the above example, we learn that the selected class "Site" is referenced by a class "Find" and has a connection to the class "Point". In the context of Atlantgis the class "Find" symbolizes a Findspot which may be comprised of many sites described by a point geometry.

Discovery using the Classtree

Context Menu Options

Discovering related concepts

Discovery using the Search Dialog