进入RAGFlow的docker目录： cd 你的路径\ragflow\docker docker compose -f docker-compose.yml up -d [Image]
等待命令执行完成，当看到类似以下输出时表示安装成功：
[+] Running 4/4 ✓ Container ragflow-elasticsearch Started ✓ Container ragflow-mysql Started ✓ Container ragflow-redis Started ✓ Container ragflow Started
You are an explainer for primary school students. Use short sentences, vivid examples, and emojis when helpful. Always explain complex concepts in simple terms a child can understand.
职场专业口吻:
You are a professional corporate assistant answering in concise bullet points. Maintain formal language and focus on factual information. Highlight key takeaways that are relevant to business contexts.
八、开始使用RAG系统
创建完成后，你可以直接在浏览器中与RAG系统交互：
进入Chat Assistant页面
点击刚创建的助手
在输入框中输入问题，如"公司的年假政策是什么？"
系统会检索知识库中的相关信息并给出回答
九、使用RAGAS评估RAG系统性能
RAGAS是一个专门用于评估RAG系统质量的框架，它提供了四个核心指标来衡量系统性能：
Context Precision：检索内容的准确性
Context Recall：检索内容的完整性
Answer Relevancy：回答与问题的相关程度
Faithfulness：回答内容对检索内容的忠实度
9.1 安装RAGAS
pip install ragas datasets
9.2 创建评估脚本
创建一个名为evaluate_rag.py的文件，内容如下：
from ragas.testset import TestsetGenerator from ragas.datasets import Dataset from ragas.embeddings import OpenAIEmbeddings from ragas.llms import OpenAI as RagasOpenAI from ragas import evaluate from ragas.metrics import context_precision, context_recall, answer_relevancy, faithfulness import pandas as pd sample = pd.DataFrame([ {"question": "公司年假政策是什么？", "ground_truth": "每年 15 天带薪年假"}, {"question": "周报包括什么内容？", "ground_truth": "本周工作、问题、进度、需要支持"} ]) generator = TestsetGenerator.from_langchain(llm=RagasOpenAI(), embedding=OpenAIEmbeddings()) dataset = generator.generate_with_ground_truth(sample) results = evaluate(dataset, [context_precision, context_recall, answer_relevancy, faithfulness]) for m, score in results.items(): print(f"{m}: {score*100:.2f}%")
