WorldDB proposes recursive world nodes, content-addressed immutability, and programmable write-time edges for ontology-aware reconciliation in agent memory, reporting 96.4% accuracy on LongMemEval-s with gains over prior graph systems.
The reconciler’s guarantees rely on handler code being a function of input state; an LLM call breaks that, and the failure mode is silent
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
WorldDB: A Vector Graph-of-Worlds Memory Engine with Ontology-Aware Write-Time Reconciliation
WorldDB proposes recursive world nodes, content-addressed immutability, and programmable write-time edges for ontology-aware reconciliation in agent memory, reporting 96.4% accuracy on LongMemEval-s with gains over prior graph systems.