We propose a four-level evidence taxonomy (Fact, Observation, Inference, Hypothesis) for knowledge graph edges that classifies relationships by the epistemological method through which they were detected. We formalize detection pipelines for each level, apply a three-axis validation reducing 32 candidate edge types to 12 defensible ones, and present a composite weight function integrating evidence level, source credibility, temporal decay, and cross-source confirmation. Includes supporting data, validation reports, and the Prioris platform source definitions.
Building similarity graph...
Analyzing shared references across papers
Loading...
Rogier Meulenaar (Sat,) studied this question.
synapsesocial.com/papers/69b79ea18166e15b153ac2e7 — DOI: https://doi.org/10.5281/zenodo.19024611
Rogier Meulenaar
Radboud University Nijmegen
Building similarity graph...
Analyzing shared references across papers
Loading...