Azure AI Search: Implementing Hybrid Search with Vectors and Keywords

Hybrid search combines traditional keyword matching with vector similarity search, delivering better results than either approach alone. Azure AI Search provides integrated hybrid search capabilities essential for RAG applications.

Setting Up Hybrid Search Index

Create an index that supports both keyword and vector search:

from azure.search.documents.indexes import SearchIndexClient
from azure.search.documents.indexes.models import (
    SearchIndex, SearchField, SearchFieldDataType,
    VectorSearch, HnswAlgorithmConfiguration, VectorSearchProfile,
    SemanticConfiguration, SemanticField, SemanticPrioritizedFields, SemanticSearch
)
from azure.identity import DefaultAzureCredential

def create_hybrid_search_index(endpoint: str, index_name: str, vector_dimensions: int = 1536) -> SearchIndex:
    """Create an index supporting hybrid search."""

    client = SearchIndexClient(endpoint, DefaultAzureCredential())

    fields = [
        SearchField(name="id", type=SearchFieldDataType.String, key=True),
        SearchField(name="content", type=SearchFieldDataType.String, searchable=True, analyzer_name="en.microsoft"),
        SearchField(name="title", type=SearchFieldDataType.String, searchable=True),
        SearchField(
            name="content_vector",
            type=SearchFieldDataType.Collection(SearchFieldDataType.Single),
            searchable=True,
            vector_search_dimensions=vector_dimensions,
            vector_search_profile_name="vector-profile"
        ),
        SearchField(name="category", type=SearchFieldDataType.String, filterable=True)
    ]

    vector_search = VectorSearch(
        algorithms=[HnswAlgorithmConfiguration(name="hnsw-config", parameters={"m": 4, "efConstruction": 400, "efSearch": 500, "metric": "cosine"})],
        profiles=[VectorSearchProfile(name="vector-profile", algorithm_configuration_name="hnsw-config")]
    )

    return client.create_or_update_index(SearchIndex(name=index_name, fields=fields, vector_search=vector_search))

Executing Hybrid Queries

Combine keyword and vector search with semantic reranking:

from azure.search.documents import SearchClient
from azure.search.documents.models import VectorizedQuery
from openai import AzureOpenAI

class HybridSearchService:
    def __init__(self, search_endpoint: str, index_name: str, openai_client: AzureOpenAI):
        self.search_client = SearchClient(endpoint=search_endpoint, index_name=index_name, credential=DefaultAzureCredential())
        self.openai_client = openai_client

    def get_embedding(self, text: str) -> list[float]:
        """Generate embedding for search query."""
        response = self.openai_client.embeddings.create(model="text-embedding-3-small", input=text)
        return response.data[0].embedding

    def hybrid_search(self, query: str, top_k: int = 10, category_filter: str = None) -> list[dict]:
        """Execute hybrid search with semantic reranking."""

        query_vector = self.get_embedding(query)
        vector_query = VectorizedQuery(vector=query_vector, k_nearest_neighbors=top_k * 2, fields="content_vector")
        filter_expr = f"category eq '{category_filter}'" if category_filter else None

        results = self.search_client.search(
            search_text=query,
            vector_queries=[vector_query],
            filter=filter_expr,
            select=["id", "title", "content", "category"],
            top=top_k,
            query_type="semantic",
            semantic_configuration_name="semantic-config"
        )

        return [{"id": r["id"], "title": r["title"], "content": r["content"], "score": r["@search.score"]} for r in results]

Hybrid search significantly improves retrieval quality for RAG applications by leveraging both semantic understanding and exact keyword matching.