research(nightly): MinCut-Partitioned Community Graph-RAG for Agent Memory Coherence#633
Draft
ruvnet wants to merge 1 commit into
Draft
research(nightly): MinCut-Partitioned Community Graph-RAG for Agent Memory Coherence#633ruvnet wants to merge 1 commit into
ruvnet wants to merge 1 commit into
Conversation
MinCut-partitioned community graph-RAG for agent memory coherence. Three variants (FlatScan, GraphHop, CommunityRAG) implementing the CommunitySearch trait; community detection via threshold-based Union-Find on cosine similarity graph. Key results (N=2000, D=64, K=10, release build, x86_64 Linux): - CommunityRAG: 9.14µs mean, 10.8× faster than FlatScan - Recall@10: 1.000 (tight clusters) / 0.953 (overlapping) - CommunityPrecision@10: 1.000 in both experiments - 12 unit tests pass, 6 acceptance tests pass Adds: - crates/ruvector-community-rag/ (standalone workspace, no ext deps) - docs/adr/ADR-272-community-memory-retrieval.md - docs/research/nightly/2026-07-02-community-memory-retrieval/ - Cargo.toml: exclude community-rag from main workspace Co-Authored-By: claude-flow <ruv@ruv.net> Claude-Session: https://claude.ai/code/session_01C2qpMxmVG89jRv77aY9zSz
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Nightly research: community-scoped ANN retrieval via graph connectivity partitioning — a new retrieval primitive for AI agent memory that achieves 10.8× speedup over brute-force search while maintaining exact recall on clustered datasets.
FlatScan,GraphHop,CommunityRAG) implementing a sharedCommunitySearchtraitruvector-mincutKey Benchmark Results (x86_64 Linux, Rust 1.94.1)
Experiment A — Tight clusters (σ=0.40, N=2000, D=64, K=10):
CommunityRAG is 10.8× faster with zero recall or precision loss.
Experiment B — Overlapping clusters (σ=1.20): CommunityRAG achieves
community_precision = 1.000vs FlatScan's0.998at 7.4× speedup, trading 4.7% ANN recall for perfect community coherence.What's Included
crates/ruvector-community-rag/(standalone workspace, zero external deps)docs/adr/ADR-272-community-memory-retrieval.mddocs/research/nightly/2026-07-02-community-memory-retrieval/README.mddocs/research/nightly/2026-07-02-community-memory-retrieval/gist.mdResearch doc:
docs/research/nightly/2026-07-02-community-memory-retrieval/README.mdADR:
docs/adr/ADR-272-community-memory-retrieval.mdGenerated by Claude Code