Compositional knowledge architecture beats GraphRAG by +0.053 nDCG@10 in registered benchmark
letandrewcook · x · 2026-11-06
- A new preprint formalizes a 4-layer stateful "organizational cortex"—Signals, Domains, Claims, Dimensions—bridging symbolic logic and continuous geometry.
- It couples Doyle's Truth Maintenance System for dependency-driven belief revision with Gärdenfors' conceptual spaces, using versioned quality axes to position derived claims geometrically for multi-axis reranking.
- Registered benchmark results: full Stateful Reranker hits 0.481 nDCG@10 vs 0.428 for a Graph-Augmented baseline and 0.400 for Signals-Only—a +0.081 gain over passage retrieval and +0.053 over GraphRAG (p < 0.001).
Related event: Symbolic belief revision plus geometric projections boost RAG reranking(2 posts)→
More from coding & agent
- Halai Play lets anyone build and ship games via MCP with Claude or ChatGPT — aziz4ai · 2026-11-07
- Open-source tool turns Garmin health data into daily LLM agent check-ins — unixterminal · 2026-11-07
- Microsoft Ignite 2026 packs five agent-focused sessions in San Francisco — lee_stott · 2026-11-07
- Cua Driver Bench: 58% pass rate on background macOS desktop tasks vs Codex 33% — TianbaoX · 2026-11-07
- YC-backed Trepo debuts kitchen hardware that lets AI agents shop for groceries in the background — ycombinator · 2026-11-07
- Cognition ships Devin Review 2.0: 5× cheaper, 1.5× faster PR review — smtabatabaie · 2026-11-07