Elasticsearch：使用 Open AI 和 Langchain 的 RAG - Retrieval Augmented Generation （三）

这是继之前文章：

的续篇。在今天的文章中，我将详述如何使用 [#!pip3 install langchain](integrations.langchain.com/vectorstore...。这也是被推荐的使用方法。如果你还没有设置好自己的环境，请详细阅读第一篇文章。

创建应用并展示

安装包

[](integrations.langchain.com/vectorstore...。这也是被推荐的使用方法。如果你还没有设置好自己的环境，请详细阅读第一篇文章。

创建应用并展示

安装包

[](integrations.langchain.com/vectorstore...。这也是被推荐的使用方法。如果你还没有设置好自己的环境，请详细阅读第一篇文章。

创建应用并展示

安装包

javascript 复制代码

1.  from dotenv import load_dotenv
2.  from langchain.embeddings import OpenAIEmbeddings
3.  from langchain.vectorstores import ElasticsearchStore
4.  from langchain.text_splitter import CharacterTextSplitter
5.  from urllib.request import urlopen
6.  import os, json

8.  load_dotenv()

10.  openai_api_key=os.getenv('OPENAI_API_KEY')
11.  elastic_user=os.getenv('ES_USER')
12.  elastic_password=os.getenv('ES_PASSWORD')
13.  elastic_endpoint=os.getenv("ES_ENDPOINT")
14.  elastic_index_name='elasticsearch-store'

添加文档并将文档分成段落

python 复制代码

1.  with open('workplace-docs.json') as f:
2.     workplace_docs = json.load(f)

4.  print(f"Successfully loaded {len(workplace_docs)} documents")

css 复制代码

1.  metadata = []
2.  content = []

4.  for doc in workplace_docs:
5.    content.append(doc["content"])
6.    metadata.append({
7.        "name": doc["name"],
8.        "summary": doc["summary"],
9.        "rolePermissions":doc["rolePermissions"]
10.    })

12.  text_splitter = CharacterTextSplitter(chunk_size=50, chunk_overlap=0)
13.  docs = text_splitter.create_documents(content, metadatas=metadata)

把数据写入到 Elasticsearch

ini 复制代码

1.  from elasticsearch import Elasticsearch

3.  embeddings = OpenAIEmbeddings(openai_api_key=openai_api_key)

5.  url = f"https://{elastic_user}:{elastic_password}@{elastic_endpoint}:9200"
6.  connection = Elasticsearch(url, ca_certs = "./http_ca.crt", verify_certs = True)

9.  es = ElasticsearchStore.from_documents( 
10.                              docs,
11.                              embedding = embeddings, 
12.                              es_url = url, 
13.                              es_connection = connection,
14.                              index_name = elastic_index_name, 
15.                              es_user = elastic_user,
16.                              es_password = elastic_password)

展示结果

lua 复制代码

1.  def showResults(output):
2.    print("Total results: ", len(output))
3.    for index in range(len(output)):
4.      print(output[index])

Similarity / Vector Search (Approximate KNN Search) - ApproxRetrievalStrategy()

ini 复制代码

1.  query = "work from home policy"
2.  result = es.similarity_search(query=query)

4.  showResults(result)

Hybrid Search (Approximate KNN + Keyword Search) - ApproxRetrievalStrategy()

我们在 Kibana 的 Dev Tools 里打入如下的命令：

ini 复制代码

1.  embeddings = OpenAIEmbeddings(openai_api_key=openai_api_key)

4.  es = ElasticsearchStore(
5.      es_url = url,
6.      es_connection = connection,
7.      es_user=elastic_user,
8.      es_password=elastic_password,
9.      embedding=embeddings,
10.      index_name=elastic_index_name,
11.      strategy=ElasticsearchStore.ApproxRetrievalStrategy(
12.          hybrid=True
13.      )
14.  )

16.  es.similarity_search("work from home policy")

造成这个错误的原因是因为当前的 License 模式不支持 RRF。我们去 Kibana 启动当前的授权：

我们再次运行代码：

Exact KNN Search (Brute Force) - ExactRetrievalStrategy()

ini 复制代码

1.  embeddings = OpenAIEmbeddings(openai_api_key=openai_api_key)

4.  es = ElasticsearchStore(
5.      es_url = url,
6.      es_connection = connection,
7.      es_user=elastic_user,
8.      es_password=elastic_password,
9.      embedding=embeddings,
10.      index_name=elastic_index_name,
11.      strategy=ElasticsearchStore.ExactRetrievalStrategy()
12.  )

14.  es.similarity_search("work from home policy")

Index / Search Documents using ELSER - SparseVectorRetrievalStrategy()

在这个步骤中，我们需要启动 ELSER。有关 ELSER 的启动，请参阅文章 "Elasticsearch：部署 ELSER - Elastic Learned Sparse EncoderR"。

ini 复制代码

1.  embeddings = OpenAIEmbeddings(openai_api_key=openai_api_key)

3.  es = ElasticsearchStore.from_documents(
4.      docs,
5.      es_url = url,
6.      es_connection = connection,
7.      es_user=elastic_user,
8.      es_password=elastic_password,
9.      index_,
10.      strategy=ElasticsearchStore.SparseVectorRetrievalStrategy()
11.  )

13.  es.similarity_search("work from home policy")

上面代码的整个 jupyter notebook 可以在地址 github.com/liu-xiao-gu... 下载。