Path Outlines

Research, Design & Development

2018 - 2020

Marie designed and developed Path Outlines, a visualisation tool to support data producers in browsing the statements that can be produced from their Knowledge Graph. Based on coordinate views with two novel visualisations, the broken (out)lines and the path browser, it provides a path-based summary of a Knowledge Graph.

When a set is selected, path outlines of depth 1 are displayed in the Path Browser (9), and users can select other depths (7) or other sets (8). Users can filter paths by statistical feature or name (10, 13). When a single path is hovered or selected, details are available in the detail panel (11, 14).
From overview to detail. When a single path is hovered or selected, details are available in the detail panel.When an external dataset is selected, extensions of the current path in this other dataset are presented and a line points to this dataset in the context.
A path outline: for a set of entities S (red nodes) sharing a similarity criteria C (green node), a given sequence of properties p1/p2/.../pn (light blue edges) leads to a set of objects O. S and O are characterised with a set of measures M. One can see that the starting entity for which the path is missing is taken into account in the summary.
Samples extracted from Nobel and DBpedia datasets comparing 3 representations: a) a serialisation with Turtle syntax, which is meant for machines, but verbose and difficult to decipher for humans; b) an interpretation of each triple as a sentence understandable by humans to help understand the chaining mechanism; c) a visualisation as a node link diagram showing how triples are connected and can be chained.