01Multi-tier search pipeline combining FTS5, semantic cosine, HyDE, and graph 1-hop queries with RRF fusion.
02Interactive 3D WebGL, D3 hive, and Canvas adjacency matrix graph visualizations.
03Semantic fact merger to consolidate related records and reduce redundancy using LLMs.
04Auto-extracted knowledge graph with (subject, predicate, object) triples for structured memory.
0516 GitHub stars
06Multi-representation embeddings (raw, summary, keywords, questions, compressed) for comprehensive semantic search.