Fortunate Recall: Ontology-Driven Memory Lifecycle Management for Persistent Coherence in LLMs
Ansuman Mullick and Eray Tuzun (Bilkent University) classify personal facts into a behavioral ontology and apply category-specific retention policies as deterministic functions over LLM-extracted metadata, then locate through ablation which half of the design produces which gain.
Ask this paper
The lifecycle question: Current memory systems treat all personal facts identically, so stores grow without bound and retrieval precision falls. The design question is which memories persist, which are replaced, and at what rate, conditioned on the behavioral type of the fact.
The policy layer: A 10+1 behavioral ontology with category-specific policies: differential temporal decay, slot-key supersession, event-time validity, and category-aware retrieval routing, all deterministic given the extracted metadata.
Benchmark results: FR-Bank reaches 76.9% on LifecycleBench, a new 516-question temporal-disambiguation benchmark, ahead of Mem0, A-MEM, Memory-R1 and MemoryOS at 61% to 70.5%, and 75.2% on full LongMemEval-S, so lifecycle policies cost nothing on standard retrieval.
The pre-registered ablation splits the credit: Replacing the typed layer with three generic lifecycle primitives leaves correctness statistically unchanged at -1.7 pp, 95% CI [-6.0, +2.7]. Generic lifecycle metadata carries the correctness advantage; the behavioral ontology carries calibration, halving confabulation from 24.2% to 12.0% at p below 0.001.
End to end and transfer: Confabulation falls from Mem0's 45.1% to 22.4% over answered queries and from 32.2% to 13.0% over all queries, while answering more correctly (31.2% vs 18.6%). The ranking replicates on Kimi K2.5 and on BEAM (46.8% vs Mem0's 32.9% over 280 questions), with ontology benefit concentrated in contradiction resolution and saturating near seven policy clusters.
Abstract
Current LLM memory systems treat all personal facts identically, so stores grow without bound while retrieval precision degrades. The core challenge is lifecycle management: which memories should persist, which should be replaced, and at what rate, conditioned on the behavioral type of each fact. Fortunate Recall (FR) is a composable policy layer that classifies personal facts into a 10+1 behavioral ontology and applies category-specific lifecycle policies (differential temporal decay, slot-key supersession, event-time validity, and category-aware retrieval routing) as deterministic functions over LLM-extracted metadata. FR-Bank, our infrastructure-independent implementation, reaches a 76.9% pass rate on LifecycleBench, a new 516-question temporal-disambiguation benchmark, ahead of Mem0, A-MEM, Memory-R1, and MemoryOS (61% to 70.5%), and 75.2% on the full LongMemEval-S under the canonical Wu et al. judge protocol, so lifecycle policies impose no measurable cost on standard retrieval. A pre-registered ablation locates the gains: replacing the typed layer with three generic lifecycle primitives leaves correctness statistically unchanged (-1.7pp, 95% CI [-6.0, +2.7]), so the generic lifecycle metadata carries the correctness advantage, while the behavioral ontology carries calibration, halving downstream confabulation (12.0% vs 24.2%, p<0.001). End-to-end, FR-Bank cuts confabulation from Mem0's 45.1% to 22.4% over answered queries and from 32.2% to 13.0% over all queries while answering more of them correctly (31.2% vs 18.6%); the ranking replicates on the open-weight Kimi K2.5. The decomposition transfers to BEAM, an independently built benchmark: 46.8% correct vs Mem0's 32.9% over 280 questions, with the ontology's benefit concentrated in contradiction resolution and saturating near seven policy clusters. The ontology, benchmark, and code are released.