# Streaming Knowledge Compilation: Proactive Materiality-Scored Pinning for Time-Evolving LLM Wikis

arXiv: https://arxiv.org/abs/2606.09877 · HTML: https://mlsanity.com/paper/2606.09877/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in KV cache optimization, one of 12 directions in Efficient LLM inference and serving.
- Builds on 20 of its 52 references that are in the map.
- First posted on arXiv 3 Jun 2026.

## Placed in

- Efficient LLM inference and serving / [KV cache optimization](https://mlsanity.com/inference/kv-cache-optimization/)

## Builds on (most cited first, 20 of 20)

- [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://mlsanity.com/paper/1810.04805/) (2018)
- [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://mlsanity.com/paper/2005.11401/) (2020)
- [From Local to Global: A Graph RAG Approach to Query-Focused Summarization](https://mlsanity.com/paper/2404.16130/) (2024)
- [Understanding intermediate layers using linear classifier probes](https://mlsanity.com/paper/1610.01644/) (2016)
- [BloombergGPT: A Large Language Model for Finance](https://mlsanity.com/paper/2303.17564/) (2023)
- [Introduction to Online Convex Optimization](https://mlsanity.com/paper/1909.05207/) (2019)
- [RAPTOR: Recursive Abstractive Processing for Tree-Organized Retrieval](https://mlsanity.com/paper/2401.18059/) (2024)
- [FinBERT: Financial Sentiment Analysis with Pre-trained Language Models](https://mlsanity.com/paper/1908.10063/) (2019)
- [FinGPT: Open-Source Financial Large Language Models](https://mlsanity.com/paper/2306.06031/) (2023)
- [Are Language Models Actually Useful for Time Series Forecasting?](https://mlsanity.com/paper/2406.16964/) (2024)
- [Can ChatGPT Forecast Stock Price Movements? Return Predictability and Large Language Models](https://mlsanity.com/paper/2304.07619/) (2023)
- [StreamingQA: A Benchmark for Adaptation to New Knowledge over Time in Question Answering Models](https://mlsanity.com/paper/2205.11388/) (2022)
- [From News to Forecast: Integrating Event Analysis in LLM-Based Time Series Forecasting with Reflection](https://mlsanity.com/paper/2409.17515/) (2024)
- [WISE: Rethinking the Knowledge Memory for Lifelong Model Editing of Large Language Models](https://mlsanity.com/paper/2405.14768/) (2024)
- [The Wall Street Neophyte: A Zero-Shot Analysis of ChatGPT Over MultiModal Stock Movement Prediction Challenges](https://mlsanity.com/paper/2304.05351/) (2023)
- [Don't Do RAG: When Cache-Augmented Generation is All You Need for Knowledge Tasks](https://mlsanity.com/paper/2412.15605/) (2024)
- [FinAgentBench: A Benchmark Dataset for Agentic Retrieval in Financial Question Answering](https://mlsanity.com/paper/2508.14052/) (2025)
- [Fine-Tuning Large Language Models for Stock Return Prediction Using Newsflow](https://mlsanity.com/paper/2407.18103/) (2024)
- [RiskLabs: Predicting Financial Risk Using Large Language Model based on Multimodal and Multi-Sources Data](https://mlsanity.com/paper/2404.07452/) (2024)
- [WiCER: Wiki-memory Compile, Evaluate, Refine Iterative Knowledge Compilation for LLM Wiki Systems](https://mlsanity.com/paper/2605.07068/) (2026)

## Built on by (most cited first, 1 of 1)

- [Training a Knowledge Base: Supervised Structure Learning for Agent-Curated Document Stores](https://mlsanity.com/paper/2608.21829/) (2026)

## Abstract (arXiv)

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 φt(k,n)∈[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(√Tlog K) regret bound where varepsilon=mathbbE[|φt-hatφt|] is the only domain-specific quantity. We instantiate in two domains: finance, where φt is abnormal stock volatility predicted by frozen Llama 3.1 8B classification head (AUROC = 0.728 on 76K articles, strict temporal split; 1.49× higher realized forward volatility for predicted-material articles); and Wikipedia, where φt 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(√Tlog K) guarantee applies to any domain where knowledge gaps can be predicted from streaming signals.