Merge pull request #35588 from yorek/dm-vector-search-update-2

removed reference to JSON and updated SQL samples

Author: prmerger-automator[bot]

docs/t-sql/functions/vector-search-transact-sql.md

@@ -4,7 +4,7 @@ description: VECTOR_SEARCH search for vectors similar to a given query vectors u
 author: yorek
 ms.author: damauri
 ms.reviewer: mikeray, randolphwest
-ms.date: 10/14/2025
+ms.date: 10/16/2025
 ms.service: sql
 ms.subservice: t-sql
 ms.topic: reference
@@ -85,7 +85,7 @@ The result set returned by the `VECTOR_SEARCH` function includes:
 
 The distance column is generated by the `VECTOR_SEARCH` function itself, while all other columns come from the table referenced in the `TABLE` argument.
 
-If you use an alias for the table in the `TABLE` argument, you must use that same alias to reference its columns in the `SELECT` statement. You can't use the alias assigned to `VECTOR_SEARCH` to reference columns from the table specified in `TABLE`. This behavior is easier to understand if you think of the result set as a JSON object that merges the output of `VECTOR_SEARCH` with the table data.
+If you use an alias for the table in the `TABLE` argument, you must use that same alias to reference its columns in the `SELECT` statement. You can't use the alias assigned to `VECTOR_SEARCH` to reference columns from the table specified in `TABLE`. This behavior is easier to understand if you think of the result set built by taking the output of `VECTOR_SEARCH` and merging it with the table data.
 
 If the table specified in the `TABLE` argument already contains a column named `distance`, the behavior will be similar to a SQL join between two tables that share a column name. In such cases, you must use table aliases to disambiguate the column references—otherwise, an error will be raised.
 
@@ -100,9 +100,9 @@ The current preview has the following limitations:
 Vector search happens before applying any predicate. Additional predicates are applied only after the most similar vectors are returned. The following sample returns the top 10 rows with embeddings most similar to the query vector `@qv`, then applies the predicate specified in the `WHERE` clause. If none of the 10 rows associated with the vectors returned by the vector search have the `accepted` column equal to 1, the result is empty.
 
 ```sql
-SELECT s.id,
-       s.title,
-       r.distance
+SELECT TOP (10) s.id,
+                s.title,
+                r.distance
 FROM VECTOR_SEARCH(
          TABLE = dbo.sessions AS s,
          COLUMN = embedding,
@@ -126,10 +126,10 @@ The following example finds the 10 most similar articles to the `Pink Floyd musi
 
 ```sql
 DECLARE @qv VECTOR(1536) = AI_GENERATE_EMBEDDING(N'Pink Floyd music style' USE MODEL Ada2Embeddings);
-SELECT
-    t.id, s.distance, t.title
-FROM
-    VECTOR_SEARCH(
+SELECT TOP (10) s.id,
+                s.title,
+                r.distance
+FROM VECTOR_SEARCH(
         TABLE = [dbo].[wikipedia_articles_embeddings] as t,
         COLUMN = [content_vector],
         SIMILAR_TO = @qv,
@@ -150,17 +150,19 @@ CREATE TABLE #t
     q NVARCHAR (MAX),
     v VECTOR(1536)
 );
+
 INSERT INTO #t
 SELECT id,
        q,
        AI_GENERATE_EMBEDDINGS(q USE MODEL Ada2Embeddings)
 FROM (VALUES
         (1, N'four legged furry animal'),
         (2, N'pink floyd music style')
-) AS S(id, q);
-SELECT t.id,
-       s.distance,
-       t.title
+    ) S(id, q);
+
+SELECT TOP (10) t.id,
+                s.distance,
+                t.title
 FROM #t AS qv
 CROSS APPLY
     VECTOR_SEARCH(
@@ -169,7 +171,7 @@ CROSS APPLY
         SIMILAR_TO = qv.v,
         METRIC = 'cosine',
         TOP_N = 10
-) AS s
+    ) AS s
 WHERE qv.id = 2
 ORDER BY s.distance;
 ```
@@ -214,23 +216,23 @@ GO
 
 -- Step 3: Create a vector index on the embedding column
 CREATE VECTOR INDEX vec_idx ON Articles(embedding)
-WITH (metric = 'cosine', type = 'diskann');
+WITH (METRIC = 'cosine', TYPE = 'diskann');
 GO
 
 -- Step 4: Perform a vector similarity search
 DECLARE @qv VECTOR(5) = '[0.3, 0.3, 0.3, 0.3, 0.3]';
-SELECT
+SELECT TOP(3)
     t.id,
     t.title,
     t.content,
     s.distance
 FROM
     VECTOR_SEARCH(
-        table = Articles AS t,
-        column = embedding,
-        similar_to = @qv,
-        metric = 'cosine',
-        top_n = 3
+        TABLE = Articles AS t,
+        COLUMN = embedding,
+        SIMILAR_TO = @qv,
+        METRIC = 'cosine',
+        TOP_N = 3
     ) AS s
 ORDER BY s.distance, t.title;
 ```