{"id":87828,"date":"2026-07-31T00:16:23","date_gmt":"2026-07-30T16:16:23","guid":{"rendered":"https:\/\/www.wsisp.com\/helps\/87828.html"},"modified":"2026-07-31T00:16:23","modified_gmt":"2026-07-30T16:16:23","slug":"ai%e6%99%ba%e8%83%bd%e5%ae%a2%e6%9c%8d%e7%b3%bb%e7%bb%9f%e5%bc%80%e5%8f%91%e5%ae%9e%e6%88%98%ef%bc%9a%e5%a4%a7%e6%a8%a1%e5%9e%8b%e5%9c%a8%e4%bc%81%e4%b8%9a%e5%ba%94%e7%94%a8%e4%b8%ad%e7%9a%84%e8%90%bd","status":"publish","type":"post","link":"https:\/\/www.wsisp.com\/helps\/87828.html","title":{"rendered":"AI\u667a\u80fd\u5ba2\u670d\u7cfb\u7edf\u5f00\u53d1\u5b9e\u6218\uff1a\u5927\u6a21\u578b\u5728\u4f01\u4e1a\u5e94\u7528\u4e2d\u7684\u843d\u5730"},"content":{"rendered":"

AI\u667a\u80fd\u5ba2\u670d\u7cfb\u7edf\u5f00\u53d1\u5b9e\u6218：\u5927\u6a21\u578b\u5728\u4f01\u4e1a\u5e94\u7528\u4e2d\u7684\u843d\u5730<\/h2>\n

\u4e00\u3001\u5f15\u8a00<\/h3>\n

\u667a\u80fd\u5ba2\u670d\u6b63\u6210\u4e3a\u5927\u6a21\u578b\u5728\u4f01\u4e1a\u573a\u666f\u4e2d\u843d\u5730\u6700\u6210\u719f\u7684\u5e94\u7528\u65b9\u5411\u4e4b\u4e00\u3002\u6839\u636eIDC\u6570\u636e，\u52302026\u5e74，\u5168\u7403\u7528\u4e8e\u5ba2\u6237\u670d\u52a1\u4e0e\u4f53\u9a8c\u4f18\u5316\u7684AI\u89e3\u51b3\u65b9\u6848\u652f\u51fa\u5c06\u8fbe\u5230480\u4ebf\u7f8e\u5143；\u5168\u7403AI\u5ba2\u670d\u5e02\u573a\u89c4\u6a21\u7ea6151\u4ebf\u7f8e\u5143，\u5e74\u590d\u5408\u589e\u901f\u8fbe25.6%\u3002<\/p>\n

\u4f20\u7edf\u5ba2\u670d\u7cfb\u7edf\u4f9d\u8d56\u89c4\u5219\u5f15\u64ce\u6216\u5173\u952e\u8bcd\u5339\u914d，\u5e38\u9762\u4e34\u4e09\u5927\u75db\u70b9：\u54cd\u5e94\u5ef6\u8fdf\u9ad8（\u4e32\u884cAPI\u8c03\u7528\u5bfc\u81f4\u7528\u6237\u6d41\u5931\u7387\u4e0a\u534735%）\u3001\u77e5\u8bc6\u66f4\u65b0\u6ede\u540e（\u4eba\u5de5\u7ef4\u62a4\u5468\u671f>24\u5c0f\u65f6）\u3001\u5927\u6a21\u578b\u5e7b\u89c9\u5e72\u6270（\u7f3a\u4e4f\u5b9e\u65f6\u6570\u636e\u7ea6\u675f）\u3002\u800c\u57fa\u4e8e\u5927\u6a21\u578b+RAG\u67b6\u6784\u7684\u667a\u80fd\u5ba2\u670d，\u6b63\u5728\u4ece\u6839\u672c\u4e0a\u6539\u53d8\u8fd9\u4e00\u5207\u3002<\/p>\n

\u672c\u6587\u5c06\u5b8c\u6574\u8bb0\u5f55\u4e00\u4e2a\u4f01\u4e1a\u7ea7AI\u667a\u80fd\u5ba2\u670d\u7cfb\u7edf\u7684\u5f00\u53d1\u5168\u8fc7\u7a0b\u2014\u2014\u4ece\u67b6\u6784\u8bbe\u8ba1\u5230\u4ee3\u7801\u5b9e\u73b0，\u4ece\u672c\u5730\u90e8\u7f72\u5230\u6027\u80fd\u9a8c\u8bc1，\u63d0\u4f9b\u4e00\u5957\u53ef\u76f4\u63a5\u590d\u7528\u7684\u6280\u672f\u65b9\u6848\u3002<\/p>\n

\u4e8c\u3001\u7cfb\u7edf\u67b6\u6784\u8bbe\u8ba1<\/h3>\n

2.1 \u6574\u4f53\u67b6\u6784<\/h4>\n

\u7cfb\u7edf\u91c7\u7528\u5206\u5c42\u89e3\u8026\u7684\u5fae\u670d\u52a1\u67b6\u6784，\u5206\u4e3a\u4e94\u5c42：<\/p>\n

\u7528\u6237\u5c42（\u591a\u6e20\u9053\u63a5\u5165：Web\/APP\/API\/\u5c0f\u7a0b\u5e8f）
\n \u2193
\n\u63a5\u5165\u5c42（\u534f\u8bae\u89e3\u6790\u3001WebSocket\u957f\u8fde\u63a5\u3001\u6d41\u91cf\u6cbb\u7406）
\n \u2193
\n\u5bf9\u8bdd\u7ba1\u7406\u5c42（\u610f\u56fe\u8bc6\u522b\u3001\u72b6\u6001\u8ffd\u8e2a\u3001\u4e0a\u4e0b\u6587\u8bb0\u5fc6）
\n \u2193
\nRAG\u5f15\u64ce\u5c42（\u5411\u91cf\u68c0\u7d22\u3001\u77e5\u8bc6\u53ec\u56de\u3001\u68c0\u7d22\u878d\u5408）
\n \u2193
\n\u6570\u636e\u5c42（\u5411\u91cf\u6570\u636e\u5e93\u3001\u5173\u7cfb\u6570\u636e\u5e93\u3001\u6587\u6863\u5b58\u50a8）<\/p>\n

\u6838\u5fc3\u8bbe\u8ba1\u7406\u5ff5：\u5c06\u201c\u7406\u89e3\u7528\u6237\u8bf4\u4ec0\u4e48\u201d\u548c\u201c\u627e\u5230\u6b63\u786e\u7b54\u6848\u201d\u4e24\u4e2a\u4efb\u52a1\u89e3\u8026\u2014\u2014\u524d\u8005\u4ea4\u7ed9\u5927\u6a21\u578b，\u540e\u8005\u4ea4\u7ed9RAG\u68c0\u7d22\u7cfb\u7edf\u3002<\/p>\n

2.2 \u6280\u672f\u6808\u9009\u578b<\/h4>\n\n\u7ec4\u4ef6\u9009\u578b\u65b9\u6848\u9009\u578b\u7406\u7531<\/tr>\n\n\n\n\n\n\n\n\n
\u5927\u8bed\u8a00\u6a21\u578b<\/td>\nQwen3.5:4b（Ollama\u90e8\u7f72）<\/td>\n\u5f00\u6e90\u3001\u4e2d\u6587\u4f18\u5316\u3001\u53ef\u672c\u5730\u5316\u90e8\u7f72<\/td>\n<\/tr>\n
\u5d4c\u5165\u6a21\u578b<\/td>\nBGE-Large<\/td>\n\u4e2d\u6587\u8bed\u4e49\u68c0\u7d22SOTA<\/td>\n<\/tr>\n
\u5411\u91cf\u6570\u636e\u5e93<\/td>\nChromaDB<\/td>\n\u8f7b\u91cf\u7ea7\u3001\u6613\u96c6\u6210<\/td>\n<\/tr>\n
\u5e94\u7528\u6846\u67b6<\/td>\nLangChain<\/td>\n\u6a21\u5757\u5316\u5de5\u5177\u94fe\u3001RAG\u539f\u751f\u652f\u6301<\/td>\n<\/tr>\n
Web\u6846\u67b6<\/td>\nFastAPI<\/td>\n\u9ad8\u6027\u80fd\u3001\u5f02\u6b65\u652f\u6301<\/td>\n<\/tr>\n
\u524d\u7aef\u754c\u9762<\/td>\nGradio<\/td>\n\u5feb\u901f\u539f\u578b\u3001\u4ea4\u4e92\u53cb\u597d<\/td>\n<\/tr>\n
\u90e8\u7f72\u73af\u5883<\/td>\nDocker + \u672c\u5730\u670d\u52a1\u5668<\/td>\n\u6570\u636e\u5b89\u5168\u3001\u4f4e\u6210\u672c<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

\u4e09\u3001\u6838\u5fc3\u529f\u80fd\u5b9e\u73b0<\/h3>\n

3.1 \u73af\u5883\u642d\u5efa\u4e0e\u6a21\u578b\u90e8\u7f72<\/h4>\n

\u9996\u5148\u4f7f\u7528Ollama\u5728\u672c\u5730\u90e8\u7f72\u5927\u8bed\u8a00\u6a21\u578b\u548c\u5d4c\u5165\u6a21\u578b：<\/p>\n

# \u5b89\u88c5Ollama<\/span>
\ncurl<\/span> -fsSL<\/span> https:\/\/ollama.com\/install.sh |<\/span> sh<\/span><\/p>\n

# \u62c9\u53d6\u5927\u8bed\u8a00\u6a21\u578b（\u7528\u4e8e\u751f\u6210\u56de\u7b54）<\/span>
\nollama pull qwen3.5:4b<\/p>\n

# \u62c9\u53d6\u5d4c\u5165\u6a21\u578b（\u7528\u4e8e\u6784\u5efa\u77e5\u8bc6\u5e93）<\/span>
\nollama pull bge-large<\/p>\n

\u5b89\u88c5Python\u4f9d\u8d56：<\/p>\n

pip install<\/span> langchain langchain-chroma langchain-ollama langchain-community gradio pypdf<\/p>\n

3.2 \u77e5\u8bc6\u5e93\u6784\u5efa\u4e0e\u5411\u91cf\u5316<\/h4>\n

\u77e5\u8bc6\u5e93\u662fRAG\u7cfb\u7edf\u7684\u6838\u5fc3\u8d44\u4ea7\u3002\u7cfb\u7edf\u9700\u8981\u5c06\u4ea7\u54c1\u6587\u6863\u3001FAQ\u3001\u5386\u53f2\u5bf9\u8bdd\u7b49\u975e\u7ed3\u6784\u5316\u6570\u636e\u8f6c\u5316\u4e3a\u5411\u91cf\u7d22\u5f15\u3002<\/p>\n

\u6587\u6863\u52a0\u8f7d\u4e0e\u5206\u5757：<\/p>\n

from<\/span> langchain_community.<\/span>document_loaders import<\/span> PyPDFLoader,<\/span> TextLoader
\nfrom<\/span> langchain.<\/span>text_splitter import<\/span> RecursiveCharacterTextSplitter<\/p>\n

# \u52a0\u8f7dPDF\u6587\u6863<\/span>
\nloader =<\/span> PyPDFLoader(<\/span>"knowledge_base\/product_manual.pdf"<\/span>)<\/span>
\ndocuments =<\/span> loader.<\/span>load(<\/span>)<\/span><\/p>\n

# \u6587\u672c\u5206\u5757：\u6309\u8bed\u4e49\u5212\u5206，\u5757\u5927\u5c0f512token，\u91cd\u53e0128token<\/span>
\ntext_splitter =<\/span> RecursiveCharacterTextSplitter(<\/span>
\n chunk_size=<\/span>512<\/span>,<\/span>
\n chunk_overlap=<\/span>128<\/span>,<\/span>
\n separators=<\/span>[<\/span>"\\\\n\\\\n"<\/span>,<\/span> "\\\\n"<\/span>,<\/span> "\u3002"<\/span>,<\/span> "！"<\/span>,<\/span> "？"<\/span>,<\/span> "；"<\/span>,<\/span> "，"<\/span>,<\/span> " "<\/span>,<\/span> ""<\/span>]<\/span>
\n)<\/span>
\nchunks =<\/span> text_splitter.<\/span>split_documents(<\/span>documents)<\/span>
\nprint<\/span>(<\/span>f"\u6587\u6863\u5df2\u5206\u5272\u4e3a <\/span>{<\/span>len<\/span>(<\/span>chunks)<\/span>}<\/span><\/span> \u4e2a\u6587\u672c\u5757"<\/span><\/span>)<\/span><\/p>\n

\u5411\u91cf\u5316\u4e0e\u5b58\u50a8：<\/p>\n

from<\/span> langchain_ollama import<\/span> OllamaEmbeddings
\nfrom<\/span> langchain_chroma import<\/span> Chroma<\/p>\n

# \u521d\u59cb\u5316\u5d4c\u5165\u6a21\u578b（BGE-Large）<\/span>
\nembeddings =<\/span> OllamaEmbeddings(<\/span>model=<\/span>"bge-large"<\/span>)<\/span><\/p>\n

# \u521b\u5efa\u5411\u91cf\u6570\u636e\u5e93<\/span>
\nvector_store =<\/span> Chroma.<\/span>from_documents(<\/span>
\n documents=<\/span>chunks,<\/span>
\n embedding=<\/span>embeddings,<\/span>
\n persist_directory=<\/span>".\/chroma_db"<\/span>
\n)<\/span><\/p>\n

# \u6301\u4e45\u5316\u5b58\u50a8<\/span>
\nvector_store.<\/span>persist(<\/span>)<\/span><\/p>\n

3.3 RAG\u68c0\u7d22\u589e\u5f3a\u751f\u6210\u5f15\u64ce<\/h4>\n

RAG\u7684\u6838\u5fc3\u6d41\u7a0b\u662f：\u7528\u6237\u63d0\u95ee \u2192 \u8bed\u4e49\u68c0\u7d22 \u2192 \u76f8\u5173\u6587\u6863\u53ec\u56de \u2192 \u6ce8\u5165\u4e0a\u4e0b\u6587 \u2192 \u5927\u6a21\u578b\u751f\u6210\u56de\u7b54\u3002\u76f8\u6bd4\u7eaf\u5927\u6a21\u578b\u65b9\u6848，RAG\u67b6\u6784\u7684\u77e5\u8bc6\u66f4\u65b0\u6210\u672c\u964d\u4f4e90%，\u56de\u7b54\u51c6\u786e\u7387\u63d0\u534740%\u4ee5\u4e0a\u3002<\/p>\n

\u6df7\u5408\u68c0\u7d22\u5b9e\u73b0：<\/p>\n

from<\/span> langchain_ollama import<\/span> OllamaLLM
\nfrom<\/span> langchain.<\/span>chains import<\/span> RetrievalQA
\nfrom<\/span> langchain.<\/span>prompts import<\/span> PromptTemplate<\/p>\n

# \u521d\u59cb\u5316\u5927\u8bed\u8a00\u6a21\u578b<\/span>
\nllm =<\/span> OllamaLLM(<\/span>model=<\/span>"qwen3.5:4b"<\/span>,<\/span> temperature=<\/span>0.3<\/span>)<\/span><\/p>\n

# \u521b\u5efa\u68c0\u7d22\u5668（\u8fd4\u56detop-3\u6700\u76f8\u5173\u6587\u6863）<\/span>
\nretriever =<\/span> vector_store.<\/span>as_retriever(<\/span>
\n search_type=<\/span>"similarity"<\/span>,<\/span>
\n search_kwargs=<\/span>{<\/span>"k"<\/span>:<\/span> 3<\/span>}<\/span>
\n)<\/span><\/p>\n

# \u8bbe\u8ba1\u4e25\u683c\u7684\u63d0\u793a\u8bcd\u6a21\u677f\u2014\u2014\u5f3a\u5236\u57fa\u4e8e\u77e5\u8bc6\u5e93\u56de\u7b54<\/span>
\nSTRICT_PROMPT_TEMPLATE =<\/span> """
\n\u4f60\u662f\u4e00\u4e2a\u4e13\u4e1a\u7684\u667a\u80fd\u5ba2\u670d\u52a9\u624b\u3002\u8bf7\u4e25\u683c\u9075\u5faa\u4ee5\u4e0b\u89c4\u5219：
\n1. \u53ea\u80fd\u4f7f\u7528\u4e0b\u9762\u63d0\u4f9b\u7684\u3010\u77e5\u8bc6\u5e93\u5185\u5bb9\u3011\u6765\u56de\u7b54\u95ee\u9898
\n2. \u5982\u679c\u77e5\u8bc6\u5e93\u4e2d\u6ca1\u6709\u76f8\u5173\u4fe1\u606f，\u8bf7\u56de\u7b54"\u6839\u636e\u77e5\u8bc6\u5e93\u65e0\u6cd5\u56de\u7b54\u60a8\u7684\u95ee\u9898，\u5efa\u8bae\u8f6c\u63a5\u4eba\u5de5\u5ba2\u670d"
\n3. \u4e0d\u8981\u7f16\u9020\u4efb\u4f55\u77e5\u8bc6\u5e93\u4e2d\u4e0d\u5b58\u5728\u7684\u4fe1\u606f
\n4. \u56de\u7b54\u8981\u7b80\u6d01\u3001\u51c6\u786e\u3001\u4e13\u4e1a<\/p>\n

\u3010\u77e5\u8bc6\u5e93\u5185\u5bb9\u3011：
\n{context}<\/p>\n

\u3010\u7528\u6237\u95ee\u9898\u3011：{question}<\/p>\n

\u3010\u56de\u7b54\u3011：
\n"""<\/span><\/p>\n

PROMPT =<\/span> PromptTemplate(<\/span>
\n template=<\/span>STRICT_PROMPT_TEMPLATE,<\/span>
\n input_variables=<\/span>[<\/span>"context"<\/span>,<\/span> "question"<\/span>]<\/span>
\n)<\/span><\/p>\n

# \u6784\u5efaRAG\u95ee\u7b54\u94fe<\/span>
\nqa_chain =<\/span> RetrievalQA.<\/span>from_chain_type(<\/span>
\n llm=<\/span>llm,<\/span>
\n chain_type=<\/span>"stuff"<\/span>,<\/span>
\n retriever=<\/span>retriever,<\/span>
\n chain_type_kwargs=<\/span>{<\/span>"prompt"<\/span>:<\/span> PROMPT}<\/span>
\n)<\/span><\/p>\n

# \u6d4b\u8bd5\u68c0\u7d22<\/span>
\ndef<\/span> answer_question<\/span>(<\/span>question)<\/span>:<\/span>
\n result =<\/span> qa_chain.<\/span>invoke(<\/span>{<\/span>"query"<\/span>:<\/span> question}<\/span>)<\/span>
\n return<\/span> result[<\/span>"result"<\/span>]<\/span><\/p>\n

3.4 \u5bf9\u8bdd\u72b6\u6001\u7ba1\u7406\u4e0e\u591a\u8f6e\u5bf9\u8bdd<\/h4>\n

\u591a\u8f6e\u5bf9\u8bdd\u9700\u8981\u7ef4\u62a4\u4e0a\u4e0b\u6587\u72b6\u6001\u3002\u7cfb\u7edf\u4f7f\u7528Redis\u5b58\u50a8\u4f1a\u8bdd\u72b6\u6001，\u652f\u6301\u77ed\u671f\u8bb0\u5fc6（\u5f53\u524d\u4f1a\u8bdd10\u8f6e\u5bf9\u8bdd）\u4e0e\u957f\u671f\u8bb0\u5fc6（\u7528\u6237\u753b\u50cf）\u5206\u79bb\u3002<\/p>\n

import<\/span> redis
\nimport<\/span> json
\nfrom<\/span> typing import<\/span> List,<\/span> Dict<\/p>\n

# Redis\u8fde\u63a5<\/span>
\nredis_client =<\/span> redis.<\/span>Redis(<\/span>host=<\/span>'localhost'<\/span>,<\/span> port=<\/span>6379<\/span>,<\/span> db=<\/span>0<\/span>)<\/span><\/p>\n

class<\/span> DialogManager<\/span>:<\/span>
\n def<\/span> __init__<\/span>(<\/span>self,<\/span> session_id:<\/span> str<\/span>)<\/span>:<\/span>
\n self.<\/span>session_id =<\/span> session_id
\n self.<\/span>max_history =<\/span> 10<\/span> # \u4fdd\u7559\u6700\u8fd110\u8f6e<\/span><\/p>\n

def<\/span> get_history<\/span>(<\/span>self)<\/span> –<\/span>><\/span> List[<\/span>Dict]<\/span>:<\/span>
\n """\u83b7\u53d6\u4f1a\u8bdd\u5386\u53f2"""<\/span>
\n key =<\/span> f"dialog:<\/span>{<\/span>self.<\/span>session_id}<\/span><\/span>"<\/span><\/span>
\n data =<\/span> redis_client.<\/span>get(<\/span>key)<\/span>
\n if<\/span> data:<\/span>
\n return<\/span> json.<\/span>loads(<\/span>data)<\/span>
\n return<\/span> [<\/span>]<\/span><\/p>\n

def<\/span> add_turn<\/span>(<\/span>self,<\/span> user_msg:<\/span> str<\/span>,<\/span> bot_reply:<\/span> str<\/span>)<\/span>:<\/span>
\n """\u6dfb\u52a0\u4e00\u8f6e\u5bf9\u8bdd"""<\/span>
\n key =<\/span> f"dialog:<\/span>{<\/span>self.<\/span>session_id}<\/span><\/span>"<\/span><\/span>
\n history =<\/span> self.<\/span>get_history(<\/span>)<\/span>
\n history.<\/span>append(<\/span>{<\/span>"user"<\/span>:<\/span> user_msg,<\/span> "bot"<\/span>:<\/span> bot_reply}<\/span>)<\/span>
\n # \u53ea\u4fdd\u7559\u6700\u8fd1N\u8f6e<\/span>
\n if<\/span> len<\/span>(<\/span>history)<\/span> ><\/span> self.<\/span>max_history:<\/span>
\n history =<\/span> history[<\/span>–<\/span>self.<\/span>max_history:<\/span>]<\/span>
\n redis_client.<\/span>setex(<\/span>key,<\/span> 1800<\/span>,<\/span> json.<\/span>dumps(<\/span>history)<\/span>)<\/span> # TTL=30\u5206\u949f<\/span><\/p>\n

def<\/span> build_context<\/span>(<\/span>self)<\/span> –<\/span>><\/span> str<\/span>:<\/span>
\n """\u6784\u5efa\u4e0a\u4e0b\u6587\u63d0\u793a\u8bcd"""<\/span>
\n history =<\/span> self.<\/span>get_history(<\/span>)<\/span>
\n if<\/span> not<\/span> history:<\/span>
\n return<\/span> ""<\/span>
\n context =<\/span> "\u5386\u53f2\u5bf9\u8bdd：\\\\n"<\/span>
\n for<\/span> turn in<\/span> history:<\/span>
\n context +=<\/span> f"\u7528\u6237：<\/span>{<\/span>turn[<\/span>'user'<\/span>]<\/span>}<\/span><\/span>\\\\n\u52a9\u624b：<\/span>{<\/span>turn[<\/span>'bot'<\/span>]<\/span>}<\/span><\/span>\\\\n"<\/span><\/span>
\n return<\/span> context<\/p>\n

3.5 FastAPI\u540e\u7aef\u670d\u52a1<\/h4>\n

\u5c06\u4e0a\u8ff0\u80fd\u529b\u5c01\u88c5\u4e3aRESTful API\u670d\u52a1：<\/p>\n

from<\/span> fastapi import<\/span> FastAPI,<\/span> HTTPException
\nfrom<\/span> pydantic import<\/span> BaseModel
\nfrom<\/span> typing import<\/span> Optional<\/p>\n

app =<\/span> FastAPI(<\/span>title=<\/span>"AI\u667a\u80fd\u5ba2\u670d\u7cfb\u7edf"<\/span>)<\/span><\/p>\n

class<\/span> ChatRequest<\/span>(<\/span>BaseModel)<\/span>:<\/span>
\n session_id:<\/span> str<\/span>
\n message:<\/span> str<\/span><\/p>\n

class<\/span> ChatResponse<\/span>(<\/span>BaseModel)<\/span>:<\/span>
\n session_id:<\/span> str<\/span>
\n reply:<\/span> str<\/span>
\n source_docs:<\/span> Optional[<\/span>List[<\/span>str<\/span>]<\/span>]<\/span> =<\/span> None<\/span><\/p>\n

@app.<\/span>post<\/span>(<\/span>"\/api\/chat"<\/span>,<\/span> response_model=<\/span>ChatResponse)<\/span>
\nasync<\/span> def<\/span> chat<\/span>(<\/span>request:<\/span> ChatRequest)<\/span>:<\/span>
\n # 1. \u83b7\u53d6\u5bf9\u8bdd\u4e0a\u4e0b\u6587<\/span>
\n dm =<\/span> DialogManager(<\/span>request.<\/span>session_id)<\/span>
\n context =<\/span> dm.<\/span>build_context(<\/span>)<\/span><\/p>\n

# 2. \u6784\u5efa\u5b8c\u6574\u67e5\u8be2（\u542b\u4e0a\u4e0b\u6587）<\/span>
\n full_query =<\/span> f"<\/span>{<\/span>context}<\/span><\/span>\\\\n\u7528\u6237：<\/span>{<\/span>request.<\/span>message}<\/span><\/span>"<\/span><\/span> if<\/span> context else<\/span> request.<\/span>message<\/p>\n

# 3. RAG\u68c0\u7d22+\u751f\u6210<\/span>
\n result =<\/span> qa_chain.<\/span>invoke(<\/span>{<\/span>"query"<\/span>:<\/span> full_query}<\/span>)<\/span>
\n reply =<\/span> result[<\/span>"result"<\/span>]<\/span><\/p>\n

# 4. \u4fdd\u5b58\u5bf9\u8bdd\u5386\u53f2<\/span>
\n dm.<\/span>add_turn(<\/span>request.<\/span>message,<\/span> reply)<\/span><\/p>\n

return<\/span> ChatResponse(<\/span>
\n session_id=<\/span>request.<\/span>session_id,<\/span>
\n reply=<\/span>reply,<\/span>
\n source_docs=<\/span>result.<\/span>get(<\/span>"source_documents"<\/span>,<\/span> [<\/span>]<\/span>)<\/span>
\n )<\/span><\/p>\n

@app.<\/span>get<\/span>(<\/span>"\/api\/health"<\/span>)<\/span>
\nasync<\/span> def<\/span> health_check<\/span>(<\/span>)<\/span>:<\/span>
\n return<\/span> {<\/span>"status"<\/span>:<\/span> "ok"<\/span>,<\/span> "model"<\/span>:<\/span> "qwen3.5:4b"<\/span>}<\/span><\/p>\n

# \u542f\u52a8\u670d\u52a1：uvicorn main:app –host 0.0.0.0 –port 8000<\/span><\/p>\n

3.6 Gradio Web\u754c\u9762<\/h4>\n

\u63d0\u4f9b\u53ef\u89c6\u5316\u7684\u4ea4\u4e92\u754c\u9762：<\/p>\n

import<\/span> gradio as<\/span> gr
\nimport<\/span> requests<\/p>\n

def<\/span> chat_with_agent<\/span>(<\/span>message,<\/span> history,<\/span> session_id)<\/span>:<\/span>
\n response =<\/span> requests.<\/span>post(<\/span>
\n "http:\/\/localhost:8000\/api\/chat"<\/span>,<\/span>
\n json=<\/span>{<\/span>"session_id"<\/span>:<\/span> session_id,<\/span> "message"<\/span>:<\/span> message}<\/span>
\n )<\/span>
\n if<\/span> response.<\/span>status_code ==<\/span> 200<\/span>:<\/span>
\n data =<\/span> response.<\/span>json(<\/span>)<\/span>
\n return<\/span> data[<\/span>"reply"<\/span>]<\/span>
\n return<\/span> "\u670d\u52a1\u6682\u65f6\u4e0d\u53ef\u7528，\u8bf7\u7a0d\u540e\u91cd\u8bd5"<\/span><\/p>\n

with<\/span> gr.<\/span>Blocks(<\/span>title=<\/span>"AI\u667a\u80fd\u5ba2\u670d\u7cfb\u7edf"<\/span>)<\/span> as<\/span> demo:<\/span>
\n gr.<\/span>Markdown(<\/span>"# 🤖 AI\u667a\u80fd\u5ba2\u670d\u7cfb\u7edf"<\/span>)<\/span>
\n gr.<\/span>Markdown(<\/span>"\u57fa\u4e8eRAG+\u5927\u6a21\u578b\u7684\u4f01\u4e1a\u7ea7\u667a\u80fd\u5ba2\u670d"<\/span>)<\/span><\/p>\n

session_id =<\/span> gr.<\/span>State(<\/span>lambda<\/span>:<\/span> f"session_<\/span>{<\/span>datetime.<\/span>now(<\/span>)<\/span>.<\/span>strftime(<\/span>'%Y%m%d%H%M%S'<\/span>)<\/span>}<\/span><\/span>"<\/span><\/span>)<\/span><\/p>\n

chatbot =<\/span> gr.<\/span>Chatbot(<\/span>label=<\/span>"\u5bf9\u8bdd\u7a97\u53e3"<\/span>)<\/span>
\n msg =<\/span> gr.<\/span>Textbox(<\/span>label=<\/span>"\u8f93\u5165\u60a8\u7684\u95ee\u9898"<\/span>,<\/span> placeholder=<\/span>"\u8bf7\u95ee\u6709\u4ec0\u4e48\u53ef\u4ee5\u5e2e\u52a9\u60a8？"<\/span>)<\/span><\/p>\n

def<\/span> respond<\/span>(<\/span>message,<\/span> chat_history,<\/span> session_id)<\/span>:<\/span>
\n reply =<\/span> chat_with_agent(<\/span>message,<\/span> chat_history,<\/span> session_id)<\/span>
\n chat_history.<\/span>append(<\/span>(<\/span>message,<\/span> reply)<\/span>)<\/span>
\n return<\/span> ""<\/span>,<\/span> chat_history,<\/span> session_id<\/p>\n

msg.<\/span>submit(<\/span>respond,<\/span> [<\/span>msg,<\/span> chatbot,<\/span> session_id]<\/span>,<\/span> [<\/span>msg,<\/span> chatbot,<\/span> session_id]<\/span>)<\/span>
\n gr.<\/span>Button(<\/span>"\u6e05\u7a7a\u4f1a\u8bdd"<\/span>)<\/span>.<\/span>click(<\/span>lambda<\/span>:<\/span> (<\/span>[<\/span>]<\/span>,<\/span> None<\/span>)<\/span>,<\/span> None<\/span>,<\/span> [<\/span>chatbot,<\/span> session_id]<\/span>)<\/span><\/p>\n

demo.<\/span>launch(<\/span>server_name=<\/span>"0.0.0.0"<\/span>,<\/span> server_port=<\/span>7860<\/span>)<\/span><\/p>\n

3.7 \u9ad8\u7ea7\u7279\u6027：\u591aAgent\u67b6\u6784<\/h4>\n

\u590d\u6742\u5ba2\u670d\u573a\u666f\u53ef\u5c06\u4efb\u52a1\u62c6\u89e3\u7ed9\u591a\u4e2a\u4e13\u4e1aAgent\u534f\u540c\u5904\u7406：<\/p>\n

from<\/span> langgraph.<\/span>graph import<\/span> StateGraph,<\/span> END
\nfrom<\/span> typing import<\/span> TypedDict,<\/span> Literal<\/p>\n

class<\/span> AgentState<\/span>(<\/span>TypedDict)<\/span>:<\/span>
\n query:<\/span> str<\/span>
\n intent:<\/span> str<\/span>
\n context:<\/span> str<\/span>
\n reply:<\/span> str<\/span><\/p>\n

def<\/span> classify_intent<\/span>(<\/span>state:<\/span> AgentState)<\/span> –<\/span>><\/span> AgentState:<\/span>
\n """\u610f\u56fe\u5206\u7c7b\u8282\u70b9"""<\/span>
\n # \u8c03\u7528\u8f7b\u91cf\u7ea7\u5206\u7c7b\u6a21\u578b<\/span>
\n state[<\/span>"intent"<\/span>]<\/span> =<\/span> classify(<\/span>state[<\/span>"query"<\/span>]<\/span>)<\/span>
\n return<\/span> state<\/p>\n

def<\/span> route_by_intent<\/span>(<\/span>state:<\/span> AgentState)<\/span> –<\/span>><\/span> Literal[<\/span>"product"<\/span>,<\/span> "order"<\/span>,<\/span> "complaint"<\/span>,<\/span> "human"<\/span>]<\/span>:<\/span>
\n """\u8def\u7531\u5230\u5bf9\u5e94\u7684\u4e13\u4e1aAgent"""<\/span>
\n intent_map =<\/span> {<\/span>
\n "\u4ea7\u54c1\u54a8\u8be2"<\/span>:<\/span> "product"<\/span>,<\/span>
\n "\u8ba2\u5355\u67e5\u8be2"<\/span>:<\/span> "order"<\/span>,<\/span>
\n "\u6295\u8bc9\u5efa\u8bae"<\/span>:<\/span> "complaint"<\/span>,<\/span>
\n "\u8f6c\u4eba\u5de5"<\/span>:<\/span> "human"<\/span>
\n }<\/span>
\n return<\/span> intent_map.<\/span>get(<\/span>state[<\/span>"intent"<\/span>]<\/span>,<\/span> "product"<\/span>)<\/span><\/p>\n

# \u6784\u5efa\u591aAgent\u534f\u4f5c\u56fe<\/span>
\nworkflow =<\/span> StateGraph(<\/span>AgentState)<\/span>
\nworkflow.<\/span>add_node(<\/span>"classify"<\/span>,<\/span> classify_intent)<\/span>
\nworkflow.<\/span>add_node(<\/span>"product_agent"<\/span>,<\/span> product_agent_node)<\/span>
\nworkflow.<\/span>add_node(<\/span>"order_agent"<\/span>,<\/span> order_agent_node)<\/span>
\nworkflow.<\/span>add_node(<\/span>"complaint_agent"<\/span>,<\/span> complaint_agent_node)<\/span><\/p>\n

workflow.<\/span>set_entry_point(<\/span>"classify"<\/span>)<\/span>
\nworkflow.<\/span>add_conditional_edges(<\/span>"classify"<\/span>,<\/span> route_by_intent)<\/span>
\n# … \u5404\u8282\u70b9\u8fde\u63a5\u5230END<\/span><\/p>\n

\u56db\u3001\u90e8\u7f72\u4e0e\u9a8c\u8bc1<\/h3>\n

4.1 \u542f\u52a8\u5b8c\u6574\u7cfb\u7edf<\/h4>\n

# 1. \u542f\u52a8Ollama\u670d\u52a1<\/span>
\nollama serve<\/p>\n

# 2. \u786e\u8ba4\u6a21\u578b\u5df2\u5c31\u7eea<\/span>
\nollama list
\n# \u5e94\u663e\u793a: qwen3.5:4b, bge-large<\/span><\/p>\n

# 3. \u542f\u52a8FastAPI\u540e\u7aef<\/span>
\nuvicorn main:app –host<\/span> 0.0<\/span>.0.0 –port<\/span> 8000<\/span><\/p>\n

# 4. \u542f\u52a8Gradio\u524d\u7aef<\/span>
\npython app.py
\n# \u8bbf\u95ee http:\/\/127.0.0.1:7860<\/span><\/p>\n

4.2 \u529f\u80fd\u6d4b\u8bd5<\/h4>\n

\u6d4b\u8bd5\u7528\u4f8b1：\u77e5\u8bc6\u5e93\u5185\u95ee\u9898<\/p>\n

    \n
  • \u8f93\u5165：\u201c\u4ea7\u54c1\u7684\u8d28\u4fdd\u671f\u662f\u591a\u4e45？\u201d<\/li>\n
  • \u9884\u671f：\u4ece\u77e5\u8bc6\u5e93\u68c0\u7d22\u76f8\u5173\u5185\u5bb9\u5e76\u751f\u6210\u51c6\u786e\u56de\u7b54<\/li>\n
  • \u9a8c\u8bc1：\u56de\u7b54\u5185\u5bb9\u4e0e\u77e5\u8bc6\u5e93\u539f\u6587\u4e00\u81f4，\u65e0\u5e7b\u89c9<\/li>\n<\/ul>\n

    \u6d4b\u8bd5\u7528\u4f8b2：\u77e5\u8bc6\u5e93\u5916\u95ee\u9898<\/p>\n

      \n
    • \u8f93\u5165：\u201c\u4eca\u5929\u5929\u6c14\u600e\u4e48\u6837？\u201d<\/li>\n
    • \u9884\u671f：\u7cfb\u7edf\u62d2\u7edd\u56de\u7b54，\u5efa\u8bae\u8f6c\u4eba\u5de5<\/li>\n
    • \u9a8c\u8bc1：\u56de\u590d\u201c\u6839\u636e\u77e5\u8bc6\u5e93\u65e0\u6cd5\u56de\u7b54\u60a8\u7684\u95ee\u9898\u201d<\/li>\n<\/ul>\n

      \u6d4b\u8bd5\u7528\u4f8b3：\u591a\u8f6e\u5bf9\u8bdd<\/p>\n

        \n
      • \u8f93\u51651：\u201c\u6211\u60f3\u67e5\u8be2\u8ba2\u5355\u72b6\u6001\u201d<\/li>\n
      • \u8f93\u51652：\u201c\u8ba2\u5355\u53f7\u662fORD123456\u201d<\/li>\n
      • \u9884\u671f：\u7cfb\u7edf\u80fd\u5173\u8054\u4e0a\u4e0b\u6587，\u5b8c\u6574\u7406\u89e3\u7528\u6237\u9700\u6c42<\/li>\n<\/ul>\n

        4.3 \u6027\u80fd\u6d4b\u8bd5<\/h4>\n

        # \u4f7f\u7528locust\u8fdb\u884c\u538b\u529b\u6d4b\u8bd5<\/span>
        \n# locustfile.py<\/span>
        \nfrom<\/span> locust import<\/span> HttpUser,<\/span> task,<\/span> between<\/p>\n

        class<\/span> CustomerSupportUser<\/span>(<\/span>HttpUser)<\/span>:<\/span>
        \n wait_time =<\/span> between(<\/span>0.5<\/span>,<\/span> 2<\/span>)<\/span><\/p>\n

        @task<\/span>
        \n def<\/span> ask_question<\/span>(<\/span>self)<\/span>:<\/span>
        \n self.<\/span>client.<\/span>post(<\/span>"\/api\/chat"<\/span>,<\/span> json=<\/span>{<\/span>
        \n "session_id"<\/span>:<\/span> f"test_<\/span>{<\/span>self.<\/span>id<\/span>}<\/span><\/span>"<\/span><\/span>,<\/span>
        \n "message"<\/span>:<\/span> "\u8bf7\u95ee\u4ea7\u54c1\u7684\u9000\u6362\u8d27\u653f\u7b56\u662f\u4ec0\u4e48？"<\/span>
        \n }<\/span>)<\/span><\/p>\n

        # \u8fd0\u884c：locust -f locustfile.py –host http:\/\/localhost:8000<\/span><\/p>\n

        4.4 \u8bc4\u4f30\u6307\u6807<\/h4>\n\n\u6307\u6807\u76ee\u6807\u503c\u9a8c\u8bc1\u65b9\u5f0f<\/tr>\n\n\n\n\n\n\n
        \u68c0\u7d22\u547d\u4e2d\u7387（Hit Rate）<\/td>\n\u226585%<\/td>\n\u6d4b\u8bd5\u96c6\u68c0\u7d22\u8bc4\u4f30<\/td>\n<\/tr>\n
        \u56de\u7b54\u51c6\u786e\u7387<\/td>\n\u226590%<\/td>\n\u4eba\u5de5\u6807\u6ce8\u8bc4\u4f30<\/td>\n<\/tr>\n
        \u54cd\u5e94\u5ef6\u8fdf（P95）<\/td>\n<1.5s<\/td>\n\u538b\u6d4b\u7edf\u8ba1<\/td>\n<\/tr>\n
        \u95ee\u9898\u72ec\u7acb\u89e3\u51b3\u7387<\/td>\n\u226585%<\/td>\n\u7ebf\u4e0aA\/B\u6d4b\u8bd5<\/td>\n<\/tr>\n
        \u5e7b\u89c9\u7387<\/td>\n<5%<\/td>\n\u4e8b\u5b9e\u6027\u6838\u67e5<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

        \u4e94\u3001\u6027\u80fd\u4f18\u5316\u4e0e\u751f\u4ea7\u5b9e\u8df5<\/h3>\n

        5.1 \u5206\u5c42\u7f13\u5b58\u7b56\u7565<\/h4>\n

        \u9ad8\u9891\u95ee\u9898（\u5982\u201c\u8425\u4e1a\u65f6\u95f4\u201d\u201c\u8054\u7cfb\u65b9\u5f0f\u201d）\u53ef\u901a\u8fc7Redis\u7f13\u5b58\u76f4\u63a5\u8fd4\u56de\u9884\u5b58\u7b54\u6848，\u907f\u514d\u91cd\u590d\u8c03\u7528\u5927\u6a21\u578b\u3002<\/p>\n

        from<\/span> functools import<\/span> lru_cache
        \nimport<\/span> hashlib<\/p>\n

        class<\/span> SemanticCache<\/span>:<\/span>
        \n def<\/span> __init__<\/span>(<\/span>self)<\/span>:<\/span>
        \n self.<\/span>cache =<\/span> {<\/span>}<\/span> # \u751f\u4ea7\u73af\u5883\u5e94\u4f7f\u7528Redis<\/span><\/p>\n

        def<\/span> get_cache_key<\/span>(<\/span>self,<\/span> question:<\/span> str<\/span>)<\/span> –<\/span>><\/span> str<\/span>:<\/span>
        \n return<\/span> hashlib.<\/span>md5(<\/span>question.<\/span>encode(<\/span>)<\/span>)<\/span>.<\/span>hexdigest(<\/span>)<\/span><\/p>\n

        def<\/span> get<\/span>(<\/span>self,<\/span> question:<\/span> str<\/span>)<\/span>:<\/span>
        \n key =<\/span> self.<\/span>get_cache_key(<\/span>question)<\/span>
        \n return<\/span> self.<\/span>cache.<\/span>get(<\/span>key)<\/span><\/p>\n

        def<\/span> set<\/span>(<\/span>self,<\/span> question:<\/span> str<\/span>,<\/span> answer:<\/span> str<\/span>)<\/span>:<\/span>
        \n key =<\/span> self.<\/span>get_cache_key(<\/span>question)<\/span>
        \n self.<\/span>cache[<\/span>key]<\/span> =<\/span> answer<\/p>\n

        cache =<\/span> SemanticCache(<\/span>)<\/span><\/p>\n

        def<\/span> cached_answer<\/span>(<\/span>question:<\/span> str<\/span>)<\/span> –<\/span>><\/span> Optional[<\/span>str<\/span>]<\/span>:<\/span>
        \n cached =<\/span> cache.<\/span>get(<\/span>question)<\/span>
        \n if<\/span> cached:<\/span>
        \n return<\/span> cached
        \n # \u672a\u547d\u4e2d\u5219\u8c03\u7528RAG<\/span>
        \n answer =<\/span> qa_chain.<\/span>invoke(<\/span>{<\/span>"query"<\/span>:<\/span> question}<\/span>)<\/span>
        \n cache.<\/span>set<\/span>(<\/span>question,<\/span> answer)<\/span>
        \n return<\/span> answer<\/p>\n

        5.2 \u6a21\u578b\u91cf\u5316\u4e0e\u63a8\u7406\u4f18\u5316<\/h4>\n

        \u5c06Qwen3.5-4B\u91cf\u5316\u4e3aQ4_K_M\u683c\u5f0f（\u7ea62.5GB），\u53ef\u5728\u666e\u901aCPU\u4e0a\u8fd0\u884c，\u663e\u5b58\u9700\u6c42\u4ece16GB\u964d\u81f34GB\u4ee5\u5185\u3002<\/p>\n

        # \u4f7f\u7528Ollama\u76f4\u63a5\u62c9\u53d6\u91cf\u5316\u7248\u672c<\/span>
        \nollama pull qwen3.5:4b
        \n# \u9ed8\u8ba4\u4f7f\u7528Q4_K_M\u91cf\u5316，\u5185\u5b58\u5360\u7528\u7ea63-4GB<\/span><\/p>\n

        5.3 \u751f\u4ea7\u73af\u5883\u90e8\u7f72\u6e05\u5355<\/h4>\n
      • \u5bb9\u5668\u5316\u90e8\u7f72：\u4f7f\u7528Docker\u5c01\u88c5\u6240\u6709\u4f9d\u8d56<\/li>\n
      • \u76d1\u63a7\u544a\u8b66：\u63a5\u5165Prometheus\u76d1\u63a7API\u5ef6\u8fdf\u3001\u9519\u8bef\u7387\u3001\u6a21\u578b\u8c03\u7528\u91cf<\/li>\n
      • \u65e5\u5fd7\u5ba1\u8ba1：\u8bb0\u5f55\u6240\u6709\u5bf9\u8bdd\u7528\u4e8e\u5408\u89c4\u5ba1\u67e5\u548c\u8d28\u91cf\u5206\u6790<\/li>\n
      • \u7070\u5ea6\u53d1\u5e03：\u65b0\u6a21\u578b\u7248\u672c\u5148\u5728\u5c0f\u6d41\u91cf\u7528\u6237\u4e2d\u9a8c\u8bc1<\/li>\n
      • \u6570\u636e\u5b89\u5168：\u654f\u611f\u4fe1\u606f\u8131\u654f，\u5bf9\u8bdd\u6570\u636e\u52a0\u5bc6\u5b58\u50a8<\/li>\n

        \u516d\u3001\u843d\u5730\u6848\u4f8b\u4e0e\u6548\u679c<\/h3>\n

        \u6848\u4f8b1：\u4e00\u6c7d\u4e30\u7530\u63a5\u5165\u5927\u6a21\u578b\u667a\u80fd\u5ba2\u670d\u540e，\u5728\u7ebf\u5ba2\u670d\u673a\u5668\u4eba\u72ec\u7acb\u89e3\u51b3\u7387\u4ece37%\u63d0\u5347\u81f384%，\u6708\u5747\u81ea\u52a8\u89e3\u51b3\u5ba2\u6237\u54a8\u8be21.7\u4e07\u6b21\u3002<\/p>\n

        \u6848\u4f8b2：\u67d0\u7535\u5546\u5e73\u53f0\u901a\u8fc7500\u6761\u5bf9\u8bdd\u6837\u672c\u5fae\u8c03\u6a21\u578b，\u51c6\u786e\u7387\u4ece72%\u63d0\u5347\u81f391%\u3002<\/p>\n

        \u6848\u4f8b3：\u67d0\u91d1\u878d\u4f01\u4e1a\u91c7\u7528LangChain+\u5411\u91cf\u6570\u636e\u5e93\u91cd\u6784\u5ba2\u670d\u7cfb\u7edf\u540e，\u54cd\u5e94\u901f\u5ea6\u538b\u81f3500ms\u5185，\u7efc\u5408\u6210\u672c\u4e0b\u964d80%\u3002<\/p>\n

        \u6848\u4f8b4：\u67d0\u6d88\u8d39\u91d1\u878d\u516c\u53f8\u5b8c\u6210\u4ece\u4f20\u7edf\u5ba2\u670d\u5411\u5927\u6a21\u578b\u667a\u80fd\u5ba2\u670d\u7684\u5168\u9762\u5347\u7ea7\u540e，\u5b9e\u73b0\u4e86\u201c\u65e2\u61c2\u7528\u6237，\u53c8\u61c2\u4e1a\u52a1\u201d\u7684\u667a\u80fd\u670d\u52a1\u4f53\u9a8c\u3002<\/p>\n

        \u4e03\u3001\u603b\u7ed3\u4e0e\u5c55\u671b<\/h3>\n

        \u672c\u6587\u4ece\u96f6\u5f00\u59cb\u6784\u5efa\u4e86\u4e00\u5957\u5b8c\u6574\u7684AI\u667a\u80fd\u5ba2\u670d\u7cfb\u7edf，\u6838\u5fc3\u6280\u672f\u8def\u5f84\u4e3a：\u672c\u5730\u5316\u5927\u6a21\u578b\u90e8\u7f72（Ollama+Qwen） \u2192 \u77e5\u8bc6\u5e93\u5411\u91cf\u5316（BGE+ChromaDB） \u2192 RAG\u68c0\u7d22\u589e\u5f3a\u751f\u6210（LangChain） \u2192 API\u670d\u52a1\u5c01\u88c5（FastAPI） \u2192 Web\u4ea4\u4e92\u754c\u9762（Gradio） \u3002<\/p>\n

        \u6838\u5fc3\u7ecf\u9a8c\u603b\u7ed3：<\/p>\n

      • RAG\u662f\u964d\u4f4e\u5e7b\u89c9\u7684\u5173\u952e：\u901a\u8fc7\u201c\u57fa\u4e8e\u77e5\u8bc6\u5e93\u56de\u7b54\u201d\u7684\u4e25\u683c\u7ea6\u675f，\u5c06\u5927\u6a21\u578b\u5e7b\u89c9\u7387\u4ece15%\u964d\u81f35%\u4ee5\u4e0b<\/li>\n
      • \u6df7\u5408\u68c0\u7d22\u4f18\u4e8e\u5355\u4e00\u68c0\u7d22：\u5411\u91cf\u76f8\u4f3c\u5ea6+\u5173\u952e\u8bcd\u5339\u914d\u7684\u878d\u5408\u7b56\u7565\u53ef\u663e\u8457\u63d0\u5347\u53ec\u56de\u8d28\u91cf<\/li>\n
      • \u5206\u5c42\u7f13\u5b58\u662f\u964d\u672c\u5229\u5668：\u9ad8\u9891\u95ee\u9898\u547d\u4e2d\u7f13\u5b58\u53ef\u51cf\u5c1170%\u4ee5\u4e0a\u7684\u5927\u6a21\u578b\u8c03\u7528<\/li>\n
      • \u4eceMVP\u5f00\u59cb\u8fed\u4ee3：\u5148\u7528\u5f00\u6e90\u6a21\u578b\u548c\u8f7b\u91cf\u7ea7\u5411\u91cf\u6570\u636e\u5e93\u8dd1\u901a\u6d41\u7a0b，\u518d\u9010\u6b65\u4f18\u5316<\/li>\n

        \u4e0b\u4e00\u6b65\u6f14\u8fdb\u65b9\u5411\u5305\u62ec：\u5f15\u5165\u591aAgent\u534f\u540c\u5904\u7406\u590d\u6742\u5de5\u5355\u3001\u57fa\u4e8e\u7528\u6237\u53cd\u9988\u7684\u6301\u7eed\u5b66\u4e60\u3001\u4ee5\u53ca\u8bed\u97f3\u7b49\u591a\u6a21\u6001\u4ea4\u4e92\u80fd\u529b\u7684\u96c6\u6210\u3002<\/p>\n

        \u667a\u80fd\u5ba2\u670d\u7684\u843d\u5730\u4e0d\u662f\u4e00\u6b21\u6027\u5de5\u7a0b，\u800c\u662f\u4e00\u4e2a\u6301\u7eed\u8fed\u4ee3\u7684\u8fc7\u7a0b\u2014\u2014\u4ece\u201c\u80fd\u56de\u7b54\u95ee\u9898\u201d\u5230\u201c\u51c6\u786e\u56de\u7b54\u95ee\u9898\u201d，\u518d\u5230\u201c\u8ba9\u7528\u6237\u611f\u89c9\u4e0d\u5230\u5728\u548c\u673a\u5668\u5bf9\u8bdd\u201d\u3002<\/p>\n

        \n","protected":false},"excerpt":{"rendered":"

        AI\u667a\u80fd\u5ba2\u670d\u7cfb\u7edf\u5f00\u53d1\u5b9e\u6218：\u5927\u6a21\u578b\u5728\u4f01\u4e1a\u5e94\u7528\u4e2d\u7684\u843d\u5730
        \n\u4e00\u3001\u5f15\u8a00
        \n\u667a\u80fd\u5ba2\u670d\u6b63\u6210\u4e3a\u5927\u6a21\u578b\u5728\u4f01\u4e1a\u573a\u666f\u4e2d\u843d\u5730\u6700\u6210\u719f\u7684\u5e94\u7528\u65b9\u5411\u4e4b\u4e00\u3002\u6839\u636eIDC\u6570\u636e，\u52302026\u5e74，\u5168\u7403\u7528\u4e8e\u5ba2\u6237\u670d\u52a1\u4e0e\u4f53\u9a8c\u4f18\u5316\u7684AI\u89e3\u51b3\u65b9\u6848\u652f\u51fa\u5c06\u8fbe\u5230480\u4ebf\u7f8e\u5143；\u5168\u7403AI\u5ba2\u670d\u5e02\u573a\u89c4\u6a21\u7ea6151\u4ebf\u7f8e\u5143，\u5e74\u590d\u5408\u589e\u901f\u8fbe25.6%\u3002
        \n\u4f20\u7edf\u5ba2\u670d\u7cfb\u7edf\u4f9d\u8d56\u89c4\u5219\u5f15\u64ce\u6216\u5173\u952e\u8bcd\u5339\u914d，\u5e38\u9762\u4e34\u4e09\u5927\u75db\u70b9：\u54cd\u5e94\u5ef6\u8fdf\u9ad8（\u4e32\u884cAPI\u8c03\u7528\u5bfc\u81f4\u7528\u6237\u6d41\u5931<\/p>\n","protected":false},"author":2,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[50],"topic":[],"class_list":["post-87828","post","type-post","status-publish","format-standard","hentry","category-server","tag-50"],"yoast_head":"\nAI\u667a\u80fd\u5ba2\u670d\u7cfb\u7edf\u5f00\u53d1\u5b9e\u6218\uff1a\u5927\u6a21\u578b\u5728\u4f01\u4e1a\u5e94\u7528\u4e2d\u7684\u843d\u5730 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/www.wsisp.com\/helps\/87828.html\" \/>\n<meta property=\"og:locale\" content=\"zh_CN\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"AI\u667a\u80fd\u5ba2\u670d\u7cfb\u7edf\u5f00\u53d1\u5b9e\u6218\uff1a\u5927\u6a21\u578b\u5728\u4f01\u4e1a\u5e94\u7528\u4e2d\u7684\u843d\u5730 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\" \/>\n<meta property=\"og:description\" content=\"AI\u667a\u80fd\u5ba2\u670d\u7cfb\u7edf\u5f00\u53d1\u5b9e\u6218：\u5927\u6a21\u578b\u5728\u4f01\u4e1a\u5e94\u7528\u4e2d\u7684\u843d\u5730 \u4e00\u3001\u5f15\u8a00 \u667a\u80fd\u5ba2\u670d\u6b63\u6210\u4e3a\u5927\u6a21\u578b\u5728\u4f01\u4e1a\u573a\u666f\u4e2d\u843d\u5730\u6700\u6210\u719f\u7684\u5e94\u7528\u65b9\u5411\u4e4b\u4e00\u3002\u6839\u636eIDC\u6570\u636e，\u52302026\u5e74，\u5168\u7403\u7528\u4e8e\u5ba2\u6237\u670d\u52a1\u4e0e\u4f53\u9a8c\u4f18\u5316\u7684AI\u89e3\u51b3\u65b9\u6848\u652f\u51fa\u5c06\u8fbe\u5230480\u4ebf\u7f8e\u5143；\u5168\u7403AI\u5ba2\u670d\u5e02\u573a\u89c4\u6a21\u7ea6151\u4ebf\u7f8e\u5143，\u5e74\u590d\u5408\u589e\u901f\u8fbe25.6%\u3002 \u4f20\u7edf\u5ba2\u670d\u7cfb\u7edf\u4f9d\u8d56\u89c4\u5219\u5f15\u64ce\u6216\u5173\u952e\u8bcd\u5339\u914d，\u5e38\u9762\u4e34\u4e09\u5927\u75db\u70b9：\u54cd\u5e94\u5ef6\u8fdf\u9ad8（\u4e32\u884cAPI\u8c03\u7528\u5bfc\u81f4\u7528\u6237\u6d41\u5931\" \/>\n<meta property=\"og:url\" content=\"https:\/\/www.wsisp.com\/helps\/87828.html\" \/>\n<meta property=\"og:site_name\" content=\"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\" \/>\n<meta property=\"article:published_time\" content=\"2026-07-30T16:16:23+00:00\" \/>\n<meta name=\"author\" content=\"admin\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"\u4f5c\u8005\" \/>\n\t<meta name=\"twitter:data1\" content=\"admin\" \/>\n\t<meta name=\"twitter:label2\" content=\"\u9884\u8ba1\u9605\u8bfb\u65f6\u95f4\" \/>\n\t<meta name=\"twitter:data2\" content=\"6 \u5206\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/87828.html\",\"url\":\"https:\/\/www.wsisp.com\/helps\/87828.html\",\"name\":\"AI\u667a\u80fd\u5ba2\u670d\u7cfb\u7edf\u5f00\u53d1\u5b9e\u6218\uff1a\u5927\u6a21\u578b\u5728\u4f01\u4e1a\u5e94\u7528\u4e2d\u7684\u843d\u5730 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\",\"isPartOf\":{\"@id\":\"https:\/\/www.wsisp.com\/helps\/#website\"},\"datePublished\":\"2026-07-30T16:16:23+00:00\",\"dateModified\":\"2026-07-30T16:16:23+00:00\",\"author\":{\"@id\":\"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41\"},\"breadcrumb\":{\"@id\":\"https:\/\/www.wsisp.com\/helps\/87828.html#breadcrumb\"},\"inLanguage\":\"zh-Hans\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/www.wsisp.com\/helps\/87828.html\"]}]},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/87828.html#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"\u9996\u9875\",\"item\":\"https:\/\/www.wsisp.com\/helps\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"AI\u667a\u80fd\u5ba2\u670d\u7cfb\u7edf\u5f00\u53d1\u5b9e\u6218\uff1a\u5927\u6a21\u578b\u5728\u4f01\u4e1a\u5e94\u7528\u4e2d\u7684\u843d\u5730\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/#website\",\"url\":\"https:\/\/www.wsisp.com\/helps\/\",\"name\":\"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\",\"description\":\"\u9999\u6e2f\u670d\u52a1\u5668_\u9999\u6e2f\u4e91\u670d\u52a1\u5668\u8d44\u8baf_\u670d\u52a1\u5668\u5e2e\u52a9\u6587\u6863_\u670d\u52a1\u5668\u6559\u7a0b\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/www.wsisp.com\/helps\/?s={search_term_string}\"},\"query-input\":\"required name=search_term_string\"}],\"inLanguage\":\"zh-Hans\"},{\"@type\":\"Person\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41\",\"name\":\"admin\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-Hans\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery\",\"contentUrl\":\"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery\",\"caption\":\"admin\"},\"sameAs\":[\"http:\/\/wp.wsisp.com\"],\"url\":\"https:\/\/www.wsisp.com\/helps\/author\/admin\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"AI\u667a\u80fd\u5ba2\u670d\u7cfb\u7edf\u5f00\u53d1\u5b9e\u6218\uff1a\u5927\u6a21\u578b\u5728\u4f01\u4e1a\u5e94\u7528\u4e2d\u7684\u843d\u5730 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/www.wsisp.com\/helps\/87828.html","og_locale":"zh_CN","og_type":"article","og_title":"AI\u667a\u80fd\u5ba2\u670d\u7cfb\u7edf\u5f00\u53d1\u5b9e\u6218\uff1a\u5927\u6a21\u578b\u5728\u4f01\u4e1a\u5e94\u7528\u4e2d\u7684\u843d\u5730 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","og_description":"AI\u667a\u80fd\u5ba2\u670d\u7cfb\u7edf\u5f00\u53d1\u5b9e\u6218：\u5927\u6a21\u578b\u5728\u4f01\u4e1a\u5e94\u7528\u4e2d\u7684\u843d\u5730 \u4e00\u3001\u5f15\u8a00 \u667a\u80fd\u5ba2\u670d\u6b63\u6210\u4e3a\u5927\u6a21\u578b\u5728\u4f01\u4e1a\u573a\u666f\u4e2d\u843d\u5730\u6700\u6210\u719f\u7684\u5e94\u7528\u65b9\u5411\u4e4b\u4e00\u3002\u6839\u636eIDC\u6570\u636e，\u52302026\u5e74，\u5168\u7403\u7528\u4e8e\u5ba2\u6237\u670d\u52a1\u4e0e\u4f53\u9a8c\u4f18\u5316\u7684AI\u89e3\u51b3\u65b9\u6848\u652f\u51fa\u5c06\u8fbe\u5230480\u4ebf\u7f8e\u5143；\u5168\u7403AI\u5ba2\u670d\u5e02\u573a\u89c4\u6a21\u7ea6151\u4ebf\u7f8e\u5143，\u5e74\u590d\u5408\u589e\u901f\u8fbe25.6%\u3002 \u4f20\u7edf\u5ba2\u670d\u7cfb\u7edf\u4f9d\u8d56\u89c4\u5219\u5f15\u64ce\u6216\u5173\u952e\u8bcd\u5339\u914d，\u5e38\u9762\u4e34\u4e09\u5927\u75db\u70b9：\u54cd\u5e94\u5ef6\u8fdf\u9ad8（\u4e32\u884cAPI\u8c03\u7528\u5bfc\u81f4\u7528\u6237\u6d41\u5931","og_url":"https:\/\/www.wsisp.com\/helps\/87828.html","og_site_name":"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","article_published_time":"2026-07-30T16:16:23+00:00","author":"admin","twitter_card":"summary_large_image","twitter_misc":{"\u4f5c\u8005":"admin","\u9884\u8ba1\u9605\u8bfb\u65f6\u95f4":"6 \u5206"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/www.wsisp.com\/helps\/87828.html","url":"https:\/\/www.wsisp.com\/helps\/87828.html","name":"AI\u667a\u80fd\u5ba2\u670d\u7cfb\u7edf\u5f00\u53d1\u5b9e\u6218\uff1a\u5927\u6a21\u578b\u5728\u4f01\u4e1a\u5e94\u7528\u4e2d\u7684\u843d\u5730 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","isPartOf":{"@id":"https:\/\/www.wsisp.com\/helps\/#website"},"datePublished":"2026-07-30T16:16:23+00:00","dateModified":"2026-07-30T16:16:23+00:00","author":{"@id":"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41"},"breadcrumb":{"@id":"https:\/\/www.wsisp.com\/helps\/87828.html#breadcrumb"},"inLanguage":"zh-Hans","potentialAction":[{"@type":"ReadAction","target":["https:\/\/www.wsisp.com\/helps\/87828.html"]}]},{"@type":"BreadcrumbList","@id":"https:\/\/www.wsisp.com\/helps\/87828.html#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"\u9996\u9875","item":"https:\/\/www.wsisp.com\/helps"},{"@type":"ListItem","position":2,"name":"AI\u667a\u80fd\u5ba2\u670d\u7cfb\u7edf\u5f00\u53d1\u5b9e\u6218\uff1a\u5927\u6a21\u578b\u5728\u4f01\u4e1a\u5e94\u7528\u4e2d\u7684\u843d\u5730"}]},{"@type":"WebSite","@id":"https:\/\/www.wsisp.com\/helps\/#website","url":"https:\/\/www.wsisp.com\/helps\/","name":"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","description":"\u9999\u6e2f\u670d\u52a1\u5668_\u9999\u6e2f\u4e91\u670d\u52a1\u5668\u8d44\u8baf_\u670d\u52a1\u5668\u5e2e\u52a9\u6587\u6863_\u670d\u52a1\u5668\u6559\u7a0b","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/www.wsisp.com\/helps\/?s={search_term_string}"},"query-input":"required name=search_term_string"}],"inLanguage":"zh-Hans"},{"@type":"Person","@id":"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41","name":"admin","image":{"@type":"ImageObject","inLanguage":"zh-Hans","@id":"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/image\/","url":"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery","contentUrl":"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery","caption":"admin"},"sameAs":["http:\/\/wp.wsisp.com"],"url":"https:\/\/www.wsisp.com\/helps\/author\/admin"}]}},"_links":{"self":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/posts\/87828","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/comments?post=87828"}],"version-history":[{"count":0,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/posts\/87828\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/media?parent=87828"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/categories?post=87828"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/tags?post=87828"},{"taxonomy":"topic","embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/topic?post=87828"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}