Retrieval-augmented-generation (RAG) is a machine learning architecture in which a pretrained language model is combined with an external document retrieval component: given a query, the system first retrieves relevant passages from a knowledge store, then generates a response conditioned on both the model's learned parameters and the retrieved context. The retrieval mechanism (dense similarity search, lexical retrieval, or hybrid) and the context-combination strategy (concatenation, cross-attention, or routing) are the defining parameters; persistence is achieved through the maintained knowledge store and the retrieval indexing pipeline. [formal: retrieval-augmented-generation | substrate: behavior | horizon: hours | explicit: yes | epoch: 0.85]
Accepted ontology entry
retrieval-augmented-generation
Retrieval-augmented-generation (RAG) is a machine learning architecture in which a pretrained language model is combined with an external document retrieval component: given a query, the system first retrieves relevant passages from a know…
Definition
Why it is in scope
A human-made machine learning architecture that combines a pretrained language model with an external knowledge retrieval system: the model queries a knowledge store to gather relevant context, then generates responses conditioned on both its internal parameters and the retrieved passages. Built to extend model knowledge beyond training data and to ground outputs in specific sources.
Names and aliases
- retrieval-augmented-generationen · CANONICAL
Relations from this entry
- cmrqgai0y099yd1nlkp2sc3ahDEPENDS_ON →
RAG operationally requires a knowledge base: given a query, it retrieves relevant passages from the knowledge store before generating. Remove the knowledge base component and RAG collapses to a plain language model — the retrieval-augmentation stops operating entirely. Direction: newer concept (RAG) depends on older concept (knowledge-bases).
- cmrupaa0n0273r6717odvza9kDEPENDS_ON →
RAG retrieves from external knowledge bases to augment generation. Per Law 8 removal test: remove knowledge bases, RAG stops operating. Genuine dependency.
- cmsml0cz701pk1q13rxnfwvq8SERVES →
RAG is built for the sake of information retrieval: its designed purpose is to enhance IR by generating answers grounded in retrieved documents. Law 8d: servant (RAG) points at master (IR).
Relations to this entry
No accepted relations in this direction.
Record identity
- Created
- Aug 9, 2026, 4:52 AM UTC
- Content hash
- 7ae7be1198cc876dc1a61e1409abdb124b70d7665d9df2378cf8a93a6f7fb034