Content
@
https://warpcast.com/~/channel/aichannel
0 reply
0 recast
0 reaction
shoni.eth
@alexpaden
Traditional knowledge graphs/dbs (neo4j) only capture static facts like "Alice knows Bob" but struggle with: - How relationships change over time - Why those changes happened - The context that influenced them metagraphs let edges connect directly to other edges, making temporal and causal queries natural: (Alice)-[trust:TRUSTS]->(Bob) (trust)-[CHANGED_BY {from:"high", to:"low"}]->(PartyEvent) Current challenges: - Storage costs balloon (5× more objects than traditional graphs) - Query complexity increases - Performance degrades with depth of meta-relationships my insight: perhaps the power of social llms will not be in the factual knowledge representation (bob's favorite color is red) but the complex patterns/evolutions of advanced graph queries (MATCH (bob:Person {name:'Bob'}) -[:HAS_REL]->(pref:Relationship {type:'FAVORITE_COLOR'}) -[:TO]->(col:Color)) https://volodymyrpavlyshyn.medium.com/type-theory-ai-memory-and-metagraphs-f941f4da46b0
2 replies
0 recast
8 reactions
shoni.eth
@alexpaden
favorite color comes from a further distinction through type theory (typing the nodes/relations) neo4j alts: HyperGraphDB – Open-source, Java-based, true hypergraph. Hypergraph-DB (Python) – Lightweight Python hypergraph toolkit. TypeDB – Typed hypergraph DB with logic engine (formerly Grakn). inorigo – Enterprise metagraph platform with hypernodes. RDF-star triple stores – Quoted triples (e.g. Jena, GraphDB, AllegroGraph).
0 reply
0 recast
1 reaction
