Description
Chunking long documents for RAG breaks context and returns disjointed fragments. OpenKB is an open LLM knowledge base that uses reasoning-based retrieval instead of vector similarity, with no vector DB, long-document support and native multimodality.
Its wiki pages follow Google's Open Knowledge Format, with entity pages for people, organizations and places.
Reasoning retrieval:Logic over similarity.
Long documents:Context preserved.
Multimodal:Text and images.
Entity pages:People, orgs and places.
Its wiki pages follow Google's Open Knowledge Format, with entity pages for people, organizations and places.
Features
Reasoning retrieval:Logic over similarity.
Long documents:Context preserved.
Multimodal:Text and images.
Entity pages:People, orgs and places.
