Skip to main content

Azure Cognitive Search

Azure Cognitive Search (formerly known as Azure Search) is a cloud search service that gives developers infrastructure, APIs, and tools for building a rich search experience over private, heterogeneous content in web, mobile, and enterprise applications.

Vector search is currently in public preview. It's available through the Azure portal, preview REST API and beta client libraries. More info Beta client libraries are subject to potential breaking changes, please be sure to use the SDK package version identified below. azure-search-documents==11.4.0b8

Install Azure Cognitive Search SDK

pip install azure-search-documents==11.4.0b8
pip install azure-identity

Import required libraries

import openai
import os
from langchain.embeddings import OpenAIEmbeddings
from langchain.vectorstores.azuresearch import AzureSearch

Configure OpenAI settings

Configure the OpenAI settings to use Azure OpenAI or OpenAI

os.environ["OPENAI_API_TYPE"] = "azure"
os.environ["OPENAI_API_BASE"] = "YOUR_OPENAI_ENDPOINT"
os.environ["OPENAI_API_KEY"] = "YOUR_OPENAI_API_KEY"
os.environ["OPENAI_API_VERSION"] = "2023-05-15"
model: str = "text-embedding-ada-002"

Configure vector store settings

Set up the vector store settings using environment variables:

vector_store_address: str = "YOUR_AZURE_SEARCH_ENDPOINT"
vector_store_password: str = "YOUR_AZURE_SEARCH_ADMIN_KEY"

Create embeddings and vector store instances

Create instances of the OpenAIEmbeddings and AzureSearch classes:

embeddings: OpenAIEmbeddings = OpenAIEmbeddings(deployment=model, chunk_size=1)
index_name: str = "langchain-vector-demo"
vector_store: AzureSearch = AzureSearch(
azure_search_endpoint=vector_store_address,
azure_search_key=vector_store_password,
index_name=index_name,
embedding_function=embeddings.embed_query,
)

Insert text and embeddings into vector store

Add texts and metadata from the JSON data to the vector store:

from langchain.document_loaders import TextLoader
from langchain.text_splitter import CharacterTextSplitter

loader = TextLoader("../../../state_of_the_union.txt", encoding="utf-8")

documents = loader.load()
text_splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=0)
docs = text_splitter.split_documents(documents)

vector_store.add_documents(documents=docs)

Execute a pure vector similarity search using the similarity_search() method:

# Perform a similarity search
docs = vector_store.similarity_search(
query="What did the president say about Ketanji Brown Jackson",
k=3,
search_type="similarity",
)
print(docs[0].page_content)
    Tonight. I call on the Senate to: Pass the Freedom to Vote Act. Pass the John Lewis Voting Rights Act. And while you’re at it, pass the Disclose Act so Americans can know who is funding our elections. 

Tonight, I’d like to honor someone who has dedicated his life to serve this country: Justice Stephen Breyer—an Army veteran, Constitutional scholar, and retiring Justice of the United States Supreme Court. Justice Breyer, thank you for your service.

One of the most serious constitutional responsibilities a President has is nominating someone to serve on the United States Supreme Court.

And I did that 4 days ago, when I nominated Circuit Court of Appeals Judge Ketanji Brown Jackson. One of our nation’s top legal minds, who will continue Justice Breyer’s legacy of excellence.

Perform a vector similarity search with relevance scores

Execute a pure vector similarity search using the similarity_search_with_relevance_scores() method:

docs_and_scores = vector_store.similarity_search_with_relevance_scores(query="What did the president say about Ketanji Brown Jackson", k=4, score_threshold=0.80)
from pprint import pprint
pprint(docs_and_scores)
    [(Document(page_content='Tonight. I call on the Senate to: Pass the Freedom to Vote Act. Pass the John Lewis Voting Rights Act. And while you’re at it, pass the Disclose Act so Americans can know who is funding our elections. \n\nTonight, I’d like to honor someone who has dedicated his life to serve this country: Justice Stephen Breyer—an Army veteran, Constitutional scholar, and retiring Justice of the United States Supreme Court. Justice Breyer, thank you for your service. \n\nOne of the most serious constitutional responsibilities a President has is nominating someone to serve on the United States Supreme Court. \n\nAnd I did that 4 days ago, when I nominated Circuit Court of Appeals Judge Ketanji Brown Jackson. One of our nation’s top legal minds, who will continue Justice Breyer’s legacy of excellence.', metadata={'source': 'C:\\repos\\langchain-fruocco-acs\\langchain\\docs\\extras\\modules\\state_of_the_union.txt'}),
0.8441472),
(Document(page_content='Tonight. I call on the Senate to: Pass the Freedom to Vote Act. Pass the John Lewis Voting Rights Act. And while you’re at it, pass the Disclose Act so Americans can know who is funding our elections. \n\nTonight, I’d like to honor someone who has dedicated his life to serve this country: Justice Stephen Breyer—an Army veteran, Constitutional scholar, and retiring Justice of the United States Supreme Court. Justice Breyer, thank you for your service. \n\nOne of the most serious constitutional responsibilities a President has is nominating someone to serve on the United States Supreme Court. \n\nAnd I did that 4 days ago, when I nominated Circuit Court of Appeals Judge Ketanji Brown Jackson. One of our nation’s top legal minds, who will continue Justice Breyer’s legacy of excellence.', metadata={'source': 'C:\\repos\\langchain-fruocco-acs\\langchain\\docs\\extras\\modules\\state_of_the_union.txt'}),
0.8441472),
(Document(page_content='A former top litigator in private practice. A former federal public defender. And from a family of public school educators and police officers. A consensus builder. Since she’s been nominated, she’s received a broad range of support—from the Fraternal Order of Police to former judges appointed by Democrats and Republicans. \n\nAnd if we are to advance liberty and justice, we need to secure the Border and fix the immigration system. \n\nWe can do both. At our border, we’ve installed new technology like cutting-edge scanners to better detect drug smuggling. \n\nWe’ve set up joint patrols with Mexico and Guatemala to catch more human traffickers. \n\nWe’re putting in place dedicated immigration judges so families fleeing persecution and violence can have their cases heard faster. \n\nWe’re securing commitments and supporting partners in South and Central America to host more refugees and secure their own borders.', metadata={'source': 'C:\\repos\\langchain-fruocco-acs\\langchain\\docs\\extras\\modules\\state_of_the_union.txt'}),
0.82153815),
(Document(page_content='A former top litigator in private practice. A former federal public defender. And from a family of public school educators and police officers. A consensus builder. Since she’s been nominated, she’s received a broad range of support—from the Fraternal Order of Police to former judges appointed by Democrats and Republicans. \n\nAnd if we are to advance liberty and justice, we need to secure the Border and fix the immigration system. \n\nWe can do both. At our border, we’ve installed new technology like cutting-edge scanners to better detect drug smuggling. \n\nWe’ve set up joint patrols with Mexico and Guatemala to catch more human traffickers. \n\nWe’re putting in place dedicated immigration judges so families fleeing persecution and violence can have their cases heard faster. \n\nWe’re securing commitments and supporting partners in South and Central America to host more refugees and secure their own borders.', metadata={'source': 'C:\\repos\\langchain-fruocco-acs\\langchain\\docs\\extras\\modules\\state_of_the_union.txt'}),
0.82153815)]

Execute hybrid search using the search_type or hybrid_search() method:

# Perform a hybrid search
docs = vector_store.similarity_search(
query="What did the president say about Ketanji Brown Jackson",
k=3,
search_type="hybrid"
)
print(docs[0].page_content)
    Tonight. I call on the Senate to: Pass the Freedom to Vote Act. Pass the John Lewis Voting Rights Act. And while you’re at it, pass the Disclose Act so Americans can know who is funding our elections. 

Tonight, I’d like to honor someone who has dedicated his life to serve this country: Justice Stephen Breyer—an Army veteran, Constitutional scholar, and retiring Justice of the United States Supreme Court. Justice Breyer, thank you for your service.

One of the most serious constitutional responsibilities a President has is nominating someone to serve on the United States Supreme Court.

And I did that 4 days ago, when I nominated Circuit Court of Appeals Judge Ketanji Brown Jackson. One of our nation’s top legal minds, who will continue Justice Breyer’s legacy of excellence.
# Perform a hybrid search
docs = vector_store.hybrid_search(
query="What did the president say about Ketanji Brown Jackson",
k=3
)
print(docs[0].page_content)
    Tonight. I call on the Senate to: Pass the Freedom to Vote Act. Pass the John Lewis Voting Rights Act. And while you’re at it, pass the Disclose Act so Americans can know who is funding our elections. 

Tonight, I’d like to honor someone who has dedicated his life to serve this country: Justice Stephen Breyer—an Army veteran, Constitutional scholar, and retiring Justice of the United States Supreme Court. Justice Breyer, thank you for your service.

One of the most serious constitutional responsibilities a President has is nominating someone to serve on the United States Supreme Court.

And I did that 4 days ago, when I nominated Circuit Court of Appeals Judge Ketanji Brown Jackson. One of our nation’s top legal minds, who will continue Justice Breyer’s legacy of excellence.

Create a new index with custom filterable fields

from azure.search.documents.indexes.models import (
SearchableField,
SearchField,
SearchFieldDataType,
SimpleField,
ScoringProfile,
TextWeights,
)

embeddings: OpenAIEmbeddings = OpenAIEmbeddings(deployment=model, chunk_size=1)
embedding_function = embeddings.embed_query

fields = [
SimpleField(
name="id",
type=SearchFieldDataType.String,
key=True,
filterable=True,
),
SearchableField(
name="content",
type=SearchFieldDataType.String,
searchable=True,
),
SearchField(
name="content_vector",
type=SearchFieldDataType.Collection(SearchFieldDataType.Single),
searchable=True,
vector_search_dimensions=len(embedding_function("Text")),
vector_search_configuration="default",
),
SearchableField(
name="metadata",
type=SearchFieldDataType.String,
searchable=True,
),
# Additional field to store the title
SearchableField(
name="title",
type=SearchFieldDataType.String,
searchable=True,
),
# Additional field for filtering on document source
SimpleField(
name="source",
type=SearchFieldDataType.String,
filterable=True,
),
]

index_name: str = "langchain-vector-demo-custom"

vector_store: AzureSearch = AzureSearch(
azure_search_endpoint=vector_store_address,
azure_search_key=vector_store_password,
index_name=index_name,
embedding_function=embedding_function,
fields=fields,
)

Perform a query with a custom filter

# Data in the metadata dictionary with a corresponding field in the index will be added to the index
# In this example, the metadata dictionary contains a title, a source and a random field
# The title and the source will be added to the index as separate fields, but the random won't. (as it is not defined in the fields list)
# The random field will be only stored in the metadata field
vector_store.add_texts(
["Test 1", "Test 2", "Test 3"],
[
{"title": "Title 1", "source": "A", "random": "10290"},
{"title": "Title 2", "source": "A", "random": "48392"},
{"title": "Title 3", "source": "B", "random": "32893"},
],
)
res = vector_store.similarity_search(query="Test 3 source1", k=3, search_type="hybrid")
res
    [Document(page_content='Test 3', metadata={'title': 'Title 3', 'source': 'B', 'random': '32893'}),
Document(page_content='Test 1', metadata={'title': 'Title 1', 'source': 'A', 'random': '10290'}),
Document(page_content='Test 2', metadata={'title': 'Title 2', 'source': 'A', 'random': '48392'})]
res = vector_store.similarity_search(query="Test 3 source1", k=3, search_type="hybrid", filters="source eq 'A'")
res
    [Document(page_content='Test 1', metadata={'title': 'Title 1', 'source': 'A', 'random': '10290'}),
Document(page_content='Test 2', metadata={'title': 'Title 2', 'source': 'A', 'random': '48392'})]

Create a new index with a Scoring Profile

from azure.search.documents.indexes.models import (
SearchableField,
SearchField,
SearchFieldDataType,
SimpleField,
ScoringProfile,
TextWeights,
ScoringFunction,
FreshnessScoringFunction,
FreshnessScoringParameters
)

embeddings: OpenAIEmbeddings = OpenAIEmbeddings(deployment=model, chunk_size=1)
embedding_function = embeddings.embed_query

fields = [
SimpleField(
name="id",
type=SearchFieldDataType.String,
key=True,
filterable=True,
),
SearchableField(
name="content",
type=SearchFieldDataType.String,
searchable=True,
),
SearchField(
name="content_vector",
type=SearchFieldDataType.Collection(SearchFieldDataType.Single),
searchable=True,
vector_search_dimensions=len(embedding_function("Text")),
vector_search_configuration="default",
),
SearchableField(
name="metadata",
type=SearchFieldDataType.String,
searchable=True,
),
# Additional field to store the title
SearchableField(
name="title",
type=SearchFieldDataType.String,
searchable=True,
),
# Additional field for filtering on document source
SimpleField(
name="source",
type=SearchFieldDataType.String,
filterable=True,
),
# Additional data field for last doc update
SimpleField(
name="last_update",
type=SearchFieldDataType.DateTimeOffset,
searchable=True,
filterable=True
)
]
# Adding a custom scoring profile with a freshness function
sc_name = "scoring_profile"
sc = ScoringProfile(
name=sc_name,
text_weights=TextWeights(weights={"title": 5}),
function_aggregation="sum",
functions=[
FreshnessScoringFunction(
field_name="last_update",
boost=100,
parameters=FreshnessScoringParameters(boosting_duration="P2D"),
interpolation="linear"
)
]
)

index_name = "langchain-vector-demo-custom-scoring-profile"

vector_store: AzureSearch = AzureSearch(
azure_search_endpoint=vector_store_address,
azure_search_key=vector_store_password,
index_name=index_name,
embedding_function=embeddings.embed_query,
fields=fields,
scoring_profiles = [sc],
default_scoring_profile = sc_name
)
# Adding same data with different last_update to show Scoring Profile effect
from datetime import datetime, timedelta

today = datetime.utcnow().strftime('%Y-%m-%dT%H:%M:%S-00:00')
yesterday = (datetime.utcnow() - timedelta(days=1)).strftime('%Y-%m-%dT%H:%M:%S-00:00')
one_month_ago = (datetime.utcnow() - timedelta(days=30)).strftime('%Y-%m-%dT%H:%M:%S-00:00')

vector_store.add_texts(
["Test 1", "Test 1", "Test 1"],
[
{"title": "Title 1", "source": "source1", "random": "10290", "last_update": today},
{"title": "Title 1", "source": "source1", "random": "48392", "last_update": yesterday},
{"title": "Title 1", "source": "source1", "random": "32893", "last_update": one_month_ago},
],
)
    ['NjQyNTI5ZmMtNmVkYS00Njg5LTk2ZDgtMjM3OTY4NTJkYzFj',
'M2M0MGExZjAtMjhiZC00ZDkwLThmMTgtODNlN2Y2ZDVkMTMw',
'ZmFhMDE1NzMtMjZjNS00MTFiLTk0MTEtNGRkYjgwYWQwOTI0']
res = vector_store.similarity_search(query="Test 1", k=3, search_type="similarity")
res
    [Document(page_content='Test 1', metadata={'title': 'Title 1', 'source': 'source1', 'random': '10290', 'last_update': '2023-07-13T10:47:39-00:00'}),
Document(page_content='Test 1', metadata={'title': 'Title 1', 'source': 'source1', 'random': '48392', 'last_update': '2023-07-12T10:47:39-00:00'}),
Document(page_content='Test 1', metadata={'title': 'Title 1', 'source': 'source1', 'random': '32893', 'last_update': '2023-06-13T10:47:39-00:00'})]