---
title: "LangChain"
type: entity
tags: [llm, framework, rag, document-loading]
sources: ["RAG从入门到精通系列1：基础RAG"]
last_updated: 2026-04-16
---

## Basic Information
- **Type**: LLM Application Framework
- **Source**: RAG从入门到精通系列1：基础RAG

## Definition
LangChain is a framework for building LLM applications, providing over 160 different document loaders for loading data from various sources, as well as components for building RAG pipelines.

## Key Features
- **Document Loaders**: 160+ loaders for various data sources
- **Chain Abstraction**: Link retrieval and generation components together
- **Retriever Interface**: Unified abstraction for retrieval components
- **PromptTemplate**: Template system for constructing prompts
- **Integration**: Works with various LLMs (Qwen, GPT-4, Claude, etc.) and vector databases (Qdrant, Chroma, Pinecone, etc.)

## Applications in RAG
- Loading external documents via document loaders
- Splitting documents into chunks (Splits)
- Creating retrievers from vector stores
- Chaining retrieval and generation into a unified pipeline
- Converting raw AIMessage outputs to clean string results

## Related Concepts
- [[RAG]]：LangChain is commonly used to build RAG pipelines
- [[LlamaIndex]]：Alternative framework for building LLM applications
- [[向量数据库]]：Vector stores integrated with LangChain
- [[Qdrant]]：Vector database mentioned in RAG tutorials

## Related Entities
- [[Qwen]]：LLM often used with LangChain