Reasoning with Knowledge Graph Visualizations-A Mental Models Perspective

How do people reason with knowledge graphs and how can we support these reasoning activities? When working on questions like these, we propose to take into account a central theory of cognitive science, which states that knowledge is represented in the human mind as “mental models”, i.e. as structural and procedural internal representations of external phenomena (Johnson-Laird, 1983). For the construction of such mental models, multiple pieces of information are relationally interwoven by different kinds of inferential reasoning (Goodwin and Johnson-Laird, 2005). In this cognitive theory we see a clear similarity to the conception of knowledge graphs, which also model information structurally and relationally. While knowledge graph technologies primarily aim to make large data collections (up to the global web) more “readable”, understandable, and computable for machines regarding the semantic structuring of largely language-based data (Berners-Lee et al., 2001) they do create new challenges regarding human readability. In this paper we take a closer look at knowledge graph visualizations and how they can support reasoning on knowledge graphs from a mental models perspective. We illustrate this question by means of a current H2020 research and innovation project in the cultural heritage information domain, in which we construct a knowledge graph and aim to make it visually accessible for cultural heritage experts and laypeople.