Context as Infrastructure: Knowledge Compilation for Retrieval in Agentic Systems with Harnesses
中文摘要
代理式RAG搜索工具通过简单检索提高准确性。但它们暴露出上下文问题:低效循环重复查询并浪费上下文窗口。
English Summary
Agentic RAG harnesses boost accuracy with simple retrieval. But they expose a context problem: inefficient loops waste context windows by repeating queries and rereading unused documents.
arXiv:2610.41172v1 Announce Type: cross Abstract: Retrieval-augmented generation is becoming agentic. A search harness writes queries, reads results, reformulates, and decides when to stop. This largely solves accuracy: a harness with only BM25/grep and MiniMax M3 answers 85% of BrowseComp-Plus and 91% on multi-hop FRAMES in our runs, without compiled or vector indexes. It exposes a context problem: an undisciplined loop repeats queries, rereads documents, and fills the window with unused passages. The missing layer is knowledge compilation: a configurable ingestion pipeline where LLMs enrich documents with context no chunk contains alone--navigation hierarchies with claim-evidence leaves, entity graphs, timelines, mind maps, ontologies, wikis, and memory. Compilation is not a better retriever: under a competent harness, structural artifacts gave no detectable accuracy gain on FRAMES, and a PageIndex-style outline degraded at scale. It is a supply layer. A navigation tree whose leaves are claim-evidence pairs--the mechanism we measure--cuts query tokens by 91% when evidence suffices and adds 1.9% when it does not, a 38% average reduction; this implies a 2% break-even sufficiency ra…