​

Querying in Azure AI Search

​

Query Types

{
  "search": "luxury hotel",
  "searchFields": "title,description"
}

{
  "vectorQueries": [
    {
      "vector": [...],
      "fields": "contentVector",
      "k": 50
    }
  ]
}

{
  "search": "luxury hotel",
  "vectorQueries": [
    {
      "vector": [...],
      "fields": "contentVector",
      "k": 50
    }
  ]
}

POST /knowledgebases/my-kb/retrieve
{
  "query": "complex question",
  "messageHistory": [...]
}

​

Query Components

​

Search Parameters

• search: Full-text query string
• searchFields: Fields to search (optional)
• searchMode: any (OR) or all (AND)
• queryType: simple or full Lucene

​

Filters

• filter: OData filter expression
• facets: Generate category counts
• orderby: Sort expression

​

Result Control

• select: Fields to return
• top: Maximum results
• skip: Pagination offset
• count: Include total count

​

Autocomplete and Suggestions

{
  "autocomplete": "hot",
  "suggesterName": "sg",
  "searchFields": "hotelName"
}

​

Geospatial Search

{
  "filter": "geo.distance(location, geography'POINT(-122.12 47.67)') le 10"
}

• geo.distance: Distance between points
• geo.intersects: Point within polygon

​

Advanced Query Features

​

Query Syntax

​

Simple Syntax

luxury hotel AND (spa OR pool) -airport
"ocean view"
hotel*

​

Full Lucene Syntax

title:luxury^2 description:beachfront
name:/[mh]otel/
"ocean view"~5
seatle~

​

Result Ranking

​

BM25 Scoring

• Term frequency (TF)
• Inverse document frequency (IDF)
• Field length normalization

​

Vector Similarity

• Cosine similarity (default)
• Euclidean distance
• Dot product

​

RRF (Hybrid)

score = Σ 1/(k + rank)

​

Query Example

{
  "search": "luxury beachfront hotel",
  "searchFields": "hotelName,description",
  "filter": "rating ge 4.5 and priceRange eq 'high'",
  "orderby": "rating desc",
  "select": "hotelId,hotelName,description,rating",
  "vectorQueries": [
    {
      "kind": "vector",
      "vector": [-0.009, 0.018, ...],
      "fields": "descriptionVector",
      "k": 50
    }
  ],
  "queryType": "semantic",
  "semanticConfiguration": "my-semantic-config",
  "top": 10,
  "count": true
}

​

Next Steps