Merge pull request #5877 from haileytap/vectors

[Azure Search] Remove and relocate vector concepts

Author: denrea

articles/search/includes/quickstarts/search-get-started-vector-rest.md

@@ -388,7 +388,7 @@ Key takeaways about the [Create Index](/rest/api/searchservice/indexes/create) R
 
 - Vector fields must be one of the [EDM data types used for vectors](/rest/api/searchservice/supported-data-types#edm-data-types-for-vector-fields), such as `type: Collection(Edm.Single)`. Vector fields also have `dimensions` and `vectorSearchProfile` properties.
 
-- The `vectorSearch` section is an array of approximate nearest neighbor algorithm configurations and profiles. Supported algorithms include hierarchical navigable small world and exhaustive k-nearest neighbor. For more information, see [Relevance scoring in vector search](../../vector-search-ranking.md).
+- The `vectorSearch` section is an array of Approximate Nearest Neighbor (ANN) algorithm configurations and profiles. Supported algorithms include Hierarchical Navigable Small World and exhaustive K-Nearest Neighbor. For more information, see [Relevance scoring in vector search](../../vector-search-ranking.md).
 
 - The (optional) `semantic` configuration enables reranking of search results. You can rerank results in queries of type `semantic` for string fields that are specified in the configuration. To learn more, see [Semantic ranking overview](../../semantic-search-overview.md).
 
@@ -669,7 +669,7 @@ The vector query string is semantically similar to the search string, but it inc
 
 ### Single vector search
 
-1. Formulate the request. The query is a 1536 float representation of *quintessential lodging near running trails, eateries, retail*. The query is searching `DescriptionVector` and returning k-5 results. It's using the "exhaustive" override parameter to perform a full scan of the index instead of approximate nearest neighbor. An exhaustive search is useful for small indexes.
+1. Formulate the request. The query is a 1536 float representation of *quintessential lodging near running trails, eateries, retail*. The query is searching `DescriptionVector` and returning k-5 results. It's using the "exhaustive" override parameter to perform a full scan of the index instead of ANN. An exhaustive search is useful for small indexes.
 
     ```http
     ### Run a single vector query

articles/search/search-features-list.md

@@ -36,7 +36,7 @@ The following table summarizes features by category. There's feature parity in a
 |-------------------|----------|
 | Vector indexing | Within a search index, add [vector fields](vector-search-how-to-create-index.md) to support  [**vector search**](vector-search-overview.md) scenarios. Vector fields can coexist with nonvector fields in the same search document. |
 | Vector queries | [Formulate single and multiple vector queries](vector-search-how-to-query.md). |
-| Vector search algorithms | Use [Hierarchical Navigable Small World (HNSW)](vector-search-ranking.md#when-to-use-hnsw) or [exhaustive K-Nearest Neighbors (KNN)](vector-search-ranking.md#when-to-use-exhaustive-knn) to find similar vectors in a search index. |
+| Vector search algorithms | Use [Hierarchical Navigable Small World (HNSW)](vector-search-ranking.md#about-hnsw) or [exhaustive K-Nearest Neighbors (KNN)](vector-search-ranking.md#about-exhaustive-knn) to find similar vectors in a search index. |
 | Vector filters | [Apply filters before or after query execution](vector-search-filters.md) for greater precision during information retrieval. |
 | Hybrid information retrieval | Search for concepts and keywords in a single [hybrid query request](hybrid-search-how-to-query.md). </p>[**Hybrid search**](hybrid-search-overview.md) consolidates vector and text search, with optional semantic ranking and relevance tuning for best results.|
 | Integrated data chunking and vectorization | Native data chunking through [Text Split skill](cognitive-search-skill-textsplit.md). Native vectorization through [vectorizers](vector-search-how-to-configure-vectorizer.md) and embedding skills such as [AzureOpenAIEmbeddingModel](cognitive-search-skill-azure-openai-embedding.md), [Azure AI Vision multimodal](cognitive-search-skill-vision-vectorize.md), and the [AML skill](cognitive-search-aml-skill.md) that you can use to connect to endpoints in the Azure AI Foundry model catalog. </p>[**Integrated vectorization**](vector-search-integrated-vectorization.md) provides an end-to-end indexing pipeline from source files to queries.|

articles/search/vector-search-how-to-create-index.md

@@ -9,14 +9,16 @@ ms.service: azure-ai-search
 ms.custom:
   - ignite-2024
 ms.topic: how-to
-ms.date: 06/20/2025
+ms.date: 07/07/2025
 ---
 
 # Create a vector index
 
-In Azure AI Search, you can store vectors in a search index and send vector queries for matching based on semantic similarity. A vector index is defined by an index schema that has vector fields, nonvector fields, and a vector configuration section.
+In Azure AI Search, you can use [Create or Update Index (REST API)](/rest/api/searchservice/indexes/create-or-update) to store vectors in a search index. A vector index is defined by an index schema that has vector fields, nonvector fields, and a vector configuration section.
 
-The [Create or Update Index](/rest/api/searchservice/indexes/create-or-update) REST API creates the vector index. To index vectors in Azure AI Search, follow these steps:
+When you create a vector index, you implicitly create an *embedding space* that serves as the corpus for vector queries. The embedding space consists of all vector fields populated with embeddings from the same embedding model. At query time, the system compares the vector query to the indexed vectors, returning results based on semantic similarity.
+
+To index vectors in Azure AI Search, follow these steps:
 
 > [!div class="checklist"]
 > + Start with a basic schema definition.
@@ -99,8 +101,8 @@ A vector configuration includes:
 
 [**2024-07-01**](/rest/api/searchservice/search-service-api-versions#2024-07-01) is generally available. It supports a vector configuration that has:
 
-+ Hierarchical navigable small world (HNSW) algorithm.
-+ Exhaustive k-nearest neighbor (KNN) algorithm.
++ Hierarchical Navigable Small World (HNSW) algorithm.
++ Exhaustive K-Nearest Neighbor (KNN) algorithm.
 + Scalar compression.
 + Binary compression, which is available in 2024-07-01 only and in newer Azure SDK packages.
 + Oversampling.
@@ -185,7 +187,7 @@ Be sure to have a strategy for [vectorizing your content](vector-search-how-to-g
 
    + `vectorSearch.compressions.scalarQuantizationParameters.quantizedDataType` must be set to `int8`. This is the only primitive data type supported at this time. This property is optional. Default is `int8`.
 
-   + `vectorSearch.algorithms` is either `hnsw` or `exhaustiveKnn`. These are the approximate nearest neighbors (ANN) algorithms used to organize vector content during indexing.
+   + `vectorSearch.algorithms` is either `hnsw` or `exhaustiveKnn`. These are the Approximate Nearest Neighbors (ANN) algorithms used to organize vector content during indexing.
 
    + `vectorSearch.algorithms.m` is the bi-directional link count. Default is 4. The range is 4 to 10. Lower values should return less noise in the results.
 
@@ -269,7 +271,7 @@ Preview and stable API versions support the same `vectorSearch` configurations.
 
    + `vectorSearch.compressions.scalarQuantizationParameters.quantizedDataType` must be set to `int8`. This is the only primitive data type supported at this time. This property is optional. Default is `int8`.
 
-   + `vectorSearch.algorithms.kind` is either `hnsw` or `exhaustiveKnn`. These are the approximate nearest neighbors (ANN) algorithms used to organize vector content during indexing.
+   + `vectorSearch.algorithms.kind` is either `hnsw` or `exhaustiveKnn`. These are the Approximate Nearest Neighbors (ANN) algorithms used to organize vector content during indexing.
 
    + `vectorSearch.algorithms.m` is the bi-directional link count. Default is 4. The range is 4 to 10. Lower values should return less noise in the results.
 

articles/search/vector-search-how-to-index-binary-data.md

@@ -48,7 +48,7 @@ Vector search algorithms are used to create the query navigation structures duri
 
 1. In the index schema, add a `vectorSearch` section that specifies profiles and algorithms.
 
-1. Add one or more [vector search algorithms](vector-search-ranking.md) that have a similarity metric of `hamming`. It's common to use Hierarchical Navigable Small Worlds (HNSW), but you can also use Hamming distance with exhaustive K-nearest neighbors.
+1. Add one or more [vector search algorithms](vector-search-ranking.md) that have a similarity metric of `hamming`. It's common to use Hierarchical Navigable Small Worlds (HNSW), but you can also use Hamming distance with exhaustive K-Nearest Neighbors (KNN).
 
 1. Add one or more vector profiles that specify the algorithm.
 

articles/search/vector-search-how-to-quantization.md

@@ -33,7 +33,7 @@ To use built-in quantization, follow these steps:
 
 ## Prerequisites
 
-- [Vector fields in a search index](vector-search-how-to-create-index.md), with a `vectorSearch` configuration specifying either the Hierarchical Navigable Small Worlds (HNSW) or exhaustive K-nearest neighbor (eKNN) algorithm, and a new vector profile.
+- [Vector fields in a search index](vector-search-how-to-create-index.md), with a `vectorSearch` configuration specifying either the Hierarchical Navigable Small Worlds (HNSW) or exhaustive K-Nearest Neighbor (KNN) algorithm, and a new vector profile.
 
 ## Supported quantization techniques
 
@@ -60,7 +60,7 @@ API versions determine which rescoring behavior is operational for your code. Th
 | [2024-11-01-preview](/rest/api/searchservice/indexes/create-or-update?view=rest-searchservice-2024-11-01-preview&preserve-view=true) | Scalar and binary quantization on HNSW graphs | `rescoringOptions.enableRescoring` and `rescoreStorageMethod.preserveOriginals` |
 | [2025-03-01-preview](/rest/api/searchservice/indexes/create-or-update?view=rest-searchservice-2025-03-01-preview&preserve-view=true) | Binary quantization on HNSW graphs | Previous parameter combinations are still supported but binary quantization can now be rescored if original embeddings are deleted: `rescoringOptions.enableRescoring` and `rescoringOptions.rescoreStorageMethod=discardOriginals` |
 
-Only HNSW graphs allow rescoring. Exhaustive K Nearest Neighbors (eKNN) doesn't support rescoring.
+Only HNSW graphs allow rescoring. Exhaustive KNN doesn't support rescoring.
 
 <!-- - In version 2024-11-01-preview, set `rescoringOptions.enableRescoring` and `rescoreStorageMethod.preserveOriginals`
 - In version 2025-03-01-preview, set `rescoringOptions.enableRescoring` and `rescoringOptions.rescoreStorageMethod=preserveOriginals` for scalar or binary quantization, or `rescoringOptions.enableRescoring` and `rescoringOptions.rescoreStorageMethod=discardOriginals` for binary quantization only -->
@@ -300,7 +300,7 @@ POST https://[servicename].search.windows.net/indexes?api-version=2025-03-01-pre
 
 ## Add the vector search algorithm
 
-You can use HNSW algorithm or exhaustive KNN in the 2024-11-01-preview REST API or later. For the stable version, use HNSW only. If you want rescoring, you must choose HNSW.
+You can use the HNSW or eKNN algorithm in the 2024-11-01-preview REST API or later. For the stable version, use HNSW only. If you want rescoring, you must choose HNSW.
 
    ```json
    "vectorSearch": {

articles/search/vector-search-how-to-storage-options.md

@@ -20,7 +20,7 @@ Removing storage is irreversible and requires reindexing if you want it back.
 
 ## Prerequisites
 
-- [Vector fields in a search index](vector-search-how-to-create-index.md), with a `vectorSearch` configuration specifying either the Hierarchical Navigable Small Worlds (HNSW) or exhaustive K-nearest neighbor (KNN) algorithm, and a new vector profile.
+- [Vector fields in a search index](vector-search-how-to-create-index.md), with a `vectorSearch` configuration specifying either the Hierarchical Navigable Small Worlds (HNSW) or exhaustive K-Nearest Neighbor (KNN) algorithm, and a new vector profile.
 
 ## How vector fields are stored
 
@@ -30,7 +30,7 @@ For every vector field, there are up to three copies of the vectors, each servin
 |----------|-------|------------------|
 | Source vectors received during document indexing (JSON data) | Used for incremental data refresh with `merge` or `mergeOrUpload` indexing action. Also used to return "retrievable" vectors in the query response. | `stored` property on vector fields |
 | Original full-precision vectors (binary data) | Used for internal index operations and for exhaustive KNN search in older API versions. For compressed vectors, it's also used for `preserveOriginals` rescoring on an oversampled candidate set of results from ANN search. This applies to vector fields that undergo [scalar or binary quantization](vector-search-how-to-quantization.md). | `rescoringOptions.rescoreStorageMethod` property in `vectorSearch.compressions`. |
-| Vectors in the [HNSW graph for Approximate Nearest Neighbors (ANN) search](vector-search-overview.md) (HNSW graph) or vectors for exhaustive K Nearest Neighbors (eKNN index) | Used for query execution. Consists of either full-precision vectors (when no compression is applied) or quantized vectors. | Essential. There are no parameters for removing this instance. |
+| Vectors in the [HNSW graph for Approximate Nearest Neighbors (ANN) search](vector-search-overview.md) (HNSW graph) or vectors for exhaustive K-Nearest Neighbors (eKNN index) | Used for query execution. Consists of either full-precision vectors (when no compression is applied) or quantized vectors. | Essential. There are no parameters for removing this instance. |
 
 You can set properties that permanently discard the first two instances (JSON data and binary data) from vector storage, but not the last instance.
 

articles/search/vector-search-index-size.md

@@ -24,7 +24,7 @@ For each vector field, Azure AI Search constructs an internal vector index using
 > [Vector optimization techniques](vector-search-how-to-configure-compression-storage.md) are now generally available. Use capabilities like narrow data types, scalar and binary quantization, and elimination of redundant storage to reduce your vector quota and storage quota consumption.
 
 > [!NOTE]
-> Not all algorithms consume vector index size quota. Vector quotas are established based on memory requirements of approximate nearest neighbor search. Vector fields created with the Hierarchical Navigable Small World (HNSW) algorithm need to reside in memory during query execution because of the random-access nature of graph-based traversals. Vector fields using exhaustive KNN algorithm are loaded into memory dynamically in pages during query execution, and as a result do not consume vector quota.
+> Not all algorithms consume vector index size quota. Vector quotas are established based on memory requirements of Approximate Nearest Neighbor (ANN) search. Vector fields created with the Hierarchical Navigable Small World (HNSW) algorithm need to reside in memory during query execution because of the random-access nature of graph-based traversals. Vector fields using the exhaustive K-Nearest Neighbors (KNN) algorithm are loaded into memory dynamically in pages during query execution, and as a result do not consume vector quota.
 
 ## Key points about quota and vector index size
 
@@ -178,7 +178,7 @@ The storage size of one vector is determined by its dimensionality. Multiply the
 
 ### Memory overhead from the selected algorithm  
 
-Every approximate nearest neighbor (ANN) algorithm generates extra data structures in memory to enable efficient searching. These structures consume extra space within memory.  
+Every ANN algorithm generates extra data structures in memory to enable efficient searching. These structures consume extra space within memory.  
 
 **For the HNSW algorithm, the memory overhead ranges between 1% and 20% for uncompressed float32 (Edm.Single) vectors.**