Collected sources and patterns will appear here. Add from search, explore, or the patterns library.
Map<Entity, Score> + Map<Document, List<Entity>> -> RankedList<Document>
Map scored entities back to their source documents using a document-to-entity index to rank and select documents for RAG context.
Problem it solves
Graph-based scoring yields entity-level relevance, but LLMs require coherent document context for question answering.
Consumes
Emits
The real projects this mechanism was found in. Attribution is the point — this is how the best teams actually do it.