Maree treats relational discovery not as the mechanical traversal of a graph but as an epistemic problem of selection: among innumerable formally valid connections, which relations deserve attention, and how can their relevance become intelligible? The paper’s distinctive operation is to combine symbolic structure with generative explanation through a numerical “interestingness” measure that mediates between graph topology, semantic context and user orientation. This shifts explainability away from retrospective verbal gloss and toward a ranked architecture of attention. Its experimental comparison with graph-based and knowledge-based baselines makes the argument measurable, while the reported correlation between interestingness and explanation quality suggests that interpretation can be partially formalised without being reduced to mere connectivity. The work’s larger contribution lies in showing that cultural knowledge graphs require curatorial intelligence as much as computational scale: the meaningful relation is not simply present in the dataset but produced through a calibrated encounter between structure, context and interpretation.
