aiqbench analysis
This research introduces a streaming knowledge compilation framework that uses proactive materiality scoring to maintain efficient LLM wikis in dynamic information landscapes, addressing the critical limitation of static corpus assumptions. By proving a regret bound and validating the approach across finance and Wikipedia domains, the study establishes regret analysis as a superior evaluation metric over absolute QA scores for compiled knowledge systems. The method effectively generalizes beyond specific domains, demonstrating that query-relevance surrogates can reliably predict document importance before queries arrive.
Streaming Knowledge Compilation: Proactive Materiality-Scored Pinning for Time-Evolving LLM Wikis
arXiv:2606.09877v1 Announce Type: new Abstract: LLM wiki systems compile knowledge into pre-filled KV caches for efficient inference, but assume a static corpus -- an assumption that fails whenever the underlying information landscape evolves. We formalize Streaming Knowledge Compilation: given a document stream, a fixed token budget, and future queries unknown at ingestion time, maintain a compiled wiki that minimizes cumulative regret against an offline oracle with perfect foresight. The enabling insight is a materiality signal $\phi_t(k,n)\in[0,1]$ that scores document importance for entity $k$ at time $t$, acting as a query-relevance surrogate for proactive pinning before queries arrive; we prove an $O(\sqrt{T\log K})$ regret bound where $\varepsilon=\mathbb{E}[|\phi_t-\hat\phi_t|]$ is the only domain-specific quantity. We instantiate in two domains: finance, where $\phi_t$ is abnormal stock volatility predicted by frozen Llama 3.1 8B classification head (AUROC = 0.728 on 76K articles, strict temporal split; $1.49\times$ higher realized forward volatility for predicted-material articles); and Wikipedia, where $\phi_t$ is the Abnormal Edit Ratio (AER), a cross-sectionally normalized edit velocity -- showing the same algorithm generalizes beyond the finance domain. End-to-end QA evaluation on 173 matched pairs (finance) and 119 (Wikipedia) reveals a pervasive LLM-as-judge confound on post-training knowledge, establishing that regret analysis -- not absolute QA scores -- is the reliable evaluation metric for compiled knowledge systems. Finance cumulative regret converges to -20.0 (-0.12/step); Wikipedia to +16.0 (+0.13/step), with the positive sign confirming that Wikipedia edit content is genuinely post-training -- richer context consistently improves scores (No Wiki 3.80 vs. Oracle 4.74) -- and eliminates this confound. The $O(\sqrt{T\log K})$ guarantee applies to any domain where knowledge gaps can be predicted from streaming signals.
Abstract:LLM wiki systems compile knowledge into pre-filled KV caches for efficient inference, but assume a static corpus -- an assumption that fails whenever the underlying information landscape evolves. We formalize Streaming Knowledge Compilation: given a document stream, a fixed token budget, and future queries unknown at ingestion time, maintain a compiled wiki that minimizes cumulative regret against an offline oracle with perfect foresight. The enabling insight is a materiality signal $\phi_t(k,n)\in[0,1]$ that scores document importance for entity $k$ at time $t$, acting as a query-relevance surrogate for proactive pinning before queries arrive; we prove an $O(\sqrt{T\log K})$ regret bound where $\varepsilon=\mathbb{E}[|\phi_t-\hat\phi_t|]$ is the only domain-specific quantity. We instantiate in two domains: finance, where $\phi_t$ is abnormal stock volatility predicted by frozen Llama 3.1 8B classification head (AUROC = 0.728 on 76K articles, strict temporal split; $1.49\times$ higher realized forward volatility for predicted-material articles); and Wikipedia, where $\phi_t$ is the Abnormal Edit Ratio (AER), a cross-sectionally normalized edit velocity -- showing the same algorithm generalizes beyond the finance domain. End-to-end QA evaluation on 173 matched pairs (finance) and 119 (Wikipedia) reveals a pervasive LLM-as-judge confound on post-training knowledge, establishing that regret analysis -- not absolute QA scores -- is the reliable evaluation metric for compiled knowledge systems. Finance cumulative regret converges to -20.0 (-0.12/step); Wikipedia to +16.0 (+0.13/step), with the positive sign confirming that Wikipedia edit content is genuinely post-training -- richer context consistently improves scores (No Wiki 3.80 vs. Oracle 4.74) -- and eliminates this confound. The $O(\sqrt{T\log K})$ guarantee applies to any domain where knowledge gaps can be predicted from streaming signals.
Machine Learning (cs.LG); Computational Engineering, Finance, and Science (cs.CE); Computation and Language (cs.CL)
arXiv:2606.09877 [cs.LG]
(or arXiv:2606.09877v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2606.09877 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 365 words