RAG architecture

What is RAG architecture?

A system design that retrieves relevant stored information before generating a response

In other languages

한국어RAG 아키텍처

언어 모델이 답변을 만들기 전에 외부 지식 저장소에서 관련 정보를 검색해 활용하는 설계 방식

日本語RAGアーキテクチャ

検索拡張生成とも呼ばれ、外部データを検索してから生成モデルの回答に反映する設計手法

Related Terms

• Dynamic RetrievalA process in information systems where the most relevant pieces of data are actively selected and pulled from a larger database in response to a specific query.
• Retrieval-based knowledge systemsSoftware architectures that dynamically query external data sources to augment the information available to a processing engine
• Context cachingTechnique that stores previously processed data to avoid redundant computation when an AI agent repeatedly references the same information
• Semantic cachingStorage method that retrieves cached results based on the meaning of a query rather than exact keyword matches
• Lazy LoadingA software design pattern that delays the initialization of a resource until it is required, reducing initial startup latency and resource consumption
• CQRSArchitectural pattern that separates read and update operations for a data store to improve performance and scalability
• Recency windowA method that prioritizes the most recent portion of a conversation or data stream when retrieving relevant information.
• Memory-Centric ComputingA computing design approach that places data location and memory access costs at the center of system planning.