Epistemic Robustness

This page provides a deeper look at the provenance quality of the knowledge graph, breaking it down into detailed distributions of Knowledge Level and Agent Type scores are distributed across all edges.

Together, these metrics offer a way to assess the overall strength and reliability of the knowledge graph, revealing which regions are supported by direct evidence and expert curation, and which rely on more speculative or indirect information. By exploring these patterns, we gain deeper insight into where the graph is most robust and where caution may be warranted in interpretation.

Knowledge Level

Indicates how strong or certain a statement is—ranging from direct assertions and logical entailments to statistical associations and predictions.

Higher averages reflect greater epistemic confidence; lower values indicate more speculative knowledge.

0.32
Average Knowledge Level

79.40M
edges used in calculation

Agent Type

Describes how the knowledge was generated — from direct human curation to automated text mining or computational models.

Higher averages indicate more human involvement; lower values reflect more automated or speculative sources.

0.34
Average Agent Type

79.40M
edges used in calculation