Our first shelf is a working corpus of quantitative finance and machine learning knowledge.
What is on it
Today the shelf holds 321 nodes and 424 edges: 101 papers, 37 strategies, 48 models, 29 results, 28 risks, 26 assumptions, 26 datasets, and 26 sources.
Each entry is provenance-backed and confidence-rated. A strategy cites the paper that introduced it. A dataset cites the source that provides it. A result cites the study that reported it.
Provenance and confidence come first
Every entry points at a real source. Where we could not verify a figure precisely, we say so, lower the confidence, and flag it rather than silently assert it. Nothing in the shelf is invented.
Deduplication and conflict detection surface 12 flags for human review. We never overwrite a conflicting entry silently; we surface the conflict.
Deterministic and meant to grow
The seed pipeline is deterministic and reproducible from its source content. The shelf is a starter, not a finished library, and it is built to grow.
A provenance-backed shelf is the difference between knowledge and claims.