Run a Vector Search with a Couchbase SDK

  • how-to
    +
    Using a Couchbase SDK, you can run a simple or more complex vector search against a Vector Search index.

    For more information about how the Search Service scores documents in search results, see Scoring for Search Queries.

    Prerequisites

    Choose your preferred programming language to view the applicable prerequisites for the examples on this page.

    • Go

    • Java

    • Python

    • You have the Search Service enabled on a node in your database. For more information about how to change Services on your database, see Modify a Database.

    • You have the hostname for the node in your database that’s running the Search Service.

      Go to your database settings and click Nodes to view node hostnames.

    • You have created a Vector Search index.

      For more information about how to create a Vector Search index, see Create a Vector Search Index in Quick Mode.

      You can download a sample dataset to use with the procedure or examples on this page:

      To get the best results with using the sample data with the examples in this documentation, import the sample files from the dataset into your database with the following settings:

      • Use a bucket called vector-sample.

      • Use a scope called color.

      • Use a collection called rgb for rgb.json.

      • To set your document keys, use the value of the id field from each JSON document.

      For the best results, consider using the sample Vector Search index from Create a Vector Search Index in Quick Mode.

    • You have installed the Couchbase Go SDK.

      For more information about installing and using the Couchbase Go SDK, see Start Using the Go SDK.

    • You have the Search Service enabled on a node in your database. For more information about how to change Services on your database, see Modify a Database.

    • You have the hostname for the node in your database that’s running the Search Service.

      Go to your database settings and click Nodes to view node hostnames.

    • You have created a Vector Search index.

      For more information about how to create a Vector Search index, see Create a Vector Search Index in Quick Mode.

      You can download a sample dataset to use with the procedure or examples on this page:

      To get the best results with using the sample data with the examples in this documentation, import the sample files from the dataset into your database with the following settings:

      • Use a bucket called vector-sample.

      • Use a scope called color.

      • Use a collection called rgb for rgb.json.

      • To set your document keys, use the value of the id field from each JSON document.

      For the best results, consider using the sample Vector Search index from Create a Vector Search Index in Quick Mode.

    • You have installed the Couchbase Java SDK.

      For more information about installing and using the Couchbase Java SDK, see Start Using the Java SDK.

    • You have the Search Service enabled on a node in your database. For more information about how to change Services on your database, see Modify a Database.

    • You have the hostname for the node in your database that’s running the Search Service.

      Go to your database settings and click Nodes to view node hostnames.

    • You have created a Vector Search index.

      For more information about how to create a Vector Search index, see Create a Vector Search Index in Quick Mode.

      You can download a sample dataset to use with the procedure or examples on this page:

      To get the best results with using the sample data with the examples in this documentation, import the sample files from the dataset into your database with the following settings:

      • Use a bucket called vector-sample.

      • Use a scope called color.

      • Use a collection called rgb for rgb.json.

      • To set your document keys, use the value of the id field from each JSON document.

      For the best results, consider using the sample Vector Search index from Create a Vector Search Index in Quick Mode.

    • You have installed the Couchbase Python SDK.

      For more information about installing and using the Couchbase Python SDK, see Start Using the Python SDK.

    • You have created and activated a virtual environment using venv and installed packages.

      For more information about how to set up your virtual environment, see the Python Packaging User Guide.

    • You have made a requirements.txt file with the following:

      couchbase==4.1.12
      openai==1.12.0
    • You have pulled in all dependencies for the Couchbase Python SDK by running:

      pip install -r requirements.txt

    Example: Searching for a Similar Color Vector

    The sample dataset inside rgb.json has small embedding vectors inside the colorvect_l2 field. These embedding vectors describe a color using RGB values. For example, the color red has an embedding vector of [255, 0, 0].

    The following example code searches for the color navy ([0.0, 0.0, 128.0]) in the rgb.json dataset:

    • Go

    • Java

    • Python

    package main
    
    import (
        "fmt"
        "log"
        "time"
        "os"
        "github.com/couchbase/gocb/v2"
        // Include this import if you want to run a Search query using regular Search index features.
        // "github.com/couchbase/gocb/v2/search"
        "github.com/couchbase/gocb/v2/vector"
    )
    // Make sure to change CB_USERNAME, CB_PASSWORD, and CB_HOSTNAME to the username, password, and hostname for your database.
    func main() {
        connstr := "couchbases://" + os.Getenv("CB_HOSTNAME") 
        username := os.Getenv("CB_USERNAME")
        password := os.Getenv("CB_PASSWORD")
    	// Make sure to change the bucket, and scope names to match where you stored the sample data in your database. 
        bucket_name := "vector-sample"
        scope_name := "color"
    
        cluster, err := gocb.Connect(connstr, gocb.ClusterOptions{
            Authenticator: gocb.PasswordAuthenticator{
                Username: username,
                Password: password,
            },
            SecurityConfig: gocb.SecurityConfig{
                TLSSkipVerify: true, // Disables TLS certificate verification
            },
        })
        if err != nil {
            log.Fatal(err)
        }
    
        bucket := cluster.Bucket(bucket_name)
        err = bucket.WaitUntilReady(5*time.Second, nil)
        if err != nil {
            log.Fatal(err)
        }
    
        scope := bucket.Scope(scope_name)
    
        request := gocb.SearchRequest{
            VectorSearch: vector.NewSearch(
                []*vector.Query{
    			//  You can change the RGB values {0.0, 0.0, 128.0} to search for a different color.
                vector.NewQuery("colorvect_l2", []float32{0.0, 0.0, 128.0}),
            },
            nil,
           ),
        }
    	// Change the limit value to return more results. Change the fields array to return different fields from your Search index.
        opts := &gocb.SearchOptions{Limit: 3, Fields: []string{"color"}}
    
    	// Make sure to change the index name to match your Search index. 
        matchResult, err := scope.Search("color-index", request, opts)
        if err != nil {
            log.Fatal(err)
        }
    
        for matchResult.Next() {
            row := matchResult.Row()
            docID := row.ID
            var fields interface{}
            err := row.Fields(&fields)
            if err != nil {
                log.Fatal(err)
            }
            fmt.Printf("Document ID: %s, Fields: %v\n", docID, fields)
        }
    
        if err = matchResult.Err(); err != nil {
            log.Fatal(err)
        }
    }
    import com.couchbase.client.core.error.*;
    import com.couchbase.client.java.*;
    import com.couchbase.client.java.kv.*;
    import com.couchbase.client.java.json.*;
    import com.couchbase.client.java.search.*;
    import com.couchbase.client.java.search.queries.*;
    import com.couchbase.client.java.search.result.*;
    import com.couchbase.client.java.search.vector.*;
    
    import java.time.Duration;
    import java.util.Map;
    
    public class RunVectorSearchSimpleColor {
    
        public static void main(String[] args) {
        // Make sure to change CB_USERNAME, CB_PASSWORD, and CB_HOSTNAME to the username, password, and hostname for your database.
    	String endpoint = "couchbases://" + System.getenv("CB_HOSTNAME") + "?tls_verify=none"; 
    	String username = System.getenv("CB_USERNAME");
    	String password = System.getenv("CB_PASSWORD");
        // Make sure to change the bucket, scope, collection, and index names to match where you stored the sample data in your database. 
    	String bucketName = "vector-sample";
    	String scopeName = "color";
    	String collectionName = "rgb";
    	String searchIndexName = "color-index";
    
    	try {
    	    // Connect to database/cluster with specified credentials
    	    Cluster cluster = Cluster.connect(
    		    endpoint,
    		    ClusterOptions.clusterOptions(username, password).environment(env -> {
    			    // Use the pre-configured profile below to avoid latency issues with your connection.
    			    env.applyProfile("wan-development");
    		    })
    	    );
    
    	    Bucket bucket = cluster.bucket(bucketName);
    	    bucket.waitUntilReady(Duration.ofSeconds(10));
    	    Scope scope = bucket.scope(scopeName);
    	    Collection collection = scope.collection(collectionName);
    
    	    SearchRequest request = SearchRequest
    	        .create(VectorSearch.create(
                        // Change the floats in the array to search for a different color.
                        VectorQuery.create("colorvect_l2", new float[]{ 0.0f, 0.0f, 128.0f }
                    ).numCandidates(3)));
            // Make sure to change the index name to match your Search index. 
    	    SearchResult result = scope.search("color-index", request,
            // Change the limit value to return more results. Change the value or values in fields to return different fields from your Search index.
                    SearchOptions.searchOptions().limit(3).fields("color","brightness"));
    
    	    for (SearchRow row : result.rows()) {
    	      System.out.println("Found row: " + row);
    	      System.out.println("   Fields: " + row.fieldsAs(Map.class));
    	    } 
    
    	} catch (UnambiguousTimeoutException ex) {
    	    boolean authFailure = ex.toString().contains("Authentication Failure");
    	    if (authFailure) {
    		    System.out.println("Authentication Failure Detected");
    	    } else {
    		    System.out.println("Error:");
    		    System.out.println(ex.getMessage());
    	    }
    	}
        }
    }
    #!/usr/bin/env python3
    
    import os
    import sys
    from couchbase.cluster import Cluster
    from couchbase.options import ClusterOptions
    from couchbase.auth import PasswordAuthenticator
    from couchbase.exceptions import CouchbaseException
    import couchbase.search as search
    from couchbase.options import SearchOptions
    from couchbase.vector_search import VectorQuery, VectorSearch
    
    # You can change the RGB values to search for a different color
    vector = [0.0,0.0,128.0]
    
    # Make sure to change CB_USERNAME, CB_PASSWORD, and CB_HOSTNAME to the username, password, and hostname for your database.                                                                                                     
    pa = PasswordAuthenticator(os.getenv("CB_USERNAME"), os.getenv("CB_PASSWORD"))
    cluster = Cluster("couchbases://" + os.getenv("CB_HOSTNAME") + "/?ssl=no_verify", ClusterOptions(pa))
    # Make sure to change the bucket, scope, and index names to match where you stored the sample data in your database.  
    bucket = cluster.bucket("vector-sample")
    scope = bucket.scope("color")
    search_index = "color-index"
    try:
        search_req = search.SearchRequest.create(search.MatchNoneQuery()).with_vector_search(
            VectorSearch.from_vector_query(VectorQuery('colorvect_l2', vector, num_candidates=3)))
            # Change the limit value to return more results. Change the fields array to return different fields from your Search index.
        result = scope.search(search_index, search_req, SearchOptions(limit=13,fields=["color", "id"]))
        for row in result.rows():
            print("Found row: {}".format(row))
        print("Reported total rows: {}".format(
            result.metadata().metrics().total_rows()))
    except CouchbaseException as ex:
        import traceback
        traceback.print_exc()

    Example: Semantic Search with Color Descriptions

    The following code sample requires you to have a paid subscription to the OpenAI API to generate an embedding vector from a sample text string. For more information about pricing for the OpenAI API, see OpenAI’s Pricing page for embedding models.

    The rgb.json sample data contains ready-made embedding vectors for each color’s description text. For an example of how to use a ready-made vector with Vector Search, see Run a Vector Search with the Capella UI.

    If you use the sample dataset inside rgb.json, you can use the OpenAI API to generate an embedding from any text string.

    The following code generates an embedding vector with the question What color hides everything like the night?:

    • Go

    • Java

    • Python

    package main
    
    import (
        "fmt"
        "log"
        "time"
        "os"
        "github.com/couchbase/gocb/v2"
    	// Include this import if you want to run a Search query using regular Search index features.
        // "github.com/couchbase/gocb/v2/search"
        "github.com/couchbase/gocb/v2/vector"
        "bytes"
        "encoding/json"
        "io/ioutil"
        "net/http"
    )
    
    type OpenAIResponse struct {
        Data []struct {
            Embedding []float32 `json:"embedding"`
        } `json:"data"`
    }
    
    // generateVector makes a request to OpenAI's API to get an embedding vector for the given input text.
    // Make sure to replace OPENAI_API_KEY with your own API Key.
    func generateVector(inputText string) ([]float32, error) {
        openaiAPIKey := os.Getenv("OPENAI_API_KEY")
        if openaiAPIKey == "" {
            return nil, fmt.Errorf("OPENAI_API_KEY environment variable is not set")
        }
    
        requestBody, err := json.Marshal(map[string]interface{}{
            "input":  inputText,
            "model": "text-embedding-ada-002",
        })
        if err != nil {
            return nil, fmt.Errorf("error marshaling request body: %w", err)
        }
    
        request, err := http.NewRequest("POST", "https://api.openai.com/v1/embeddings", bytes.NewBuffer(requestBody))
        if err != nil {
            return nil, fmt.Errorf("error creating request: %w", err)
        }
    
        request.Header.Set("Content-Type", "application/json")
        request.Header.Set("Authorization", "Bearer "+openaiAPIKey)
    
        client := &http.Client{}
        response, err := client.Do(request)
        if err != nil {
            return nil, fmt.Errorf("error making request: %w", err)
        }
        defer response.Body.Close()
    
        if response.StatusCode != http.StatusOK {
            bodyBytes, _ := ioutil.ReadAll(response.Body)
            return nil, fmt.Errorf("API request failed with status %d: %s", response.StatusCode, string(bodyBytes))
        }
    
        var openAIResponse OpenAIResponse
        if err := json.NewDecoder(response.Body).Decode(&openAIResponse); err != nil {
            return nil, fmt.Errorf("error decoding response: %w", err)
        }
    
        if len(openAIResponse.Data) == 0 || len(openAIResponse.Data[0].Embedding) == 0 {
            return nil, fmt.Errorf("no embedding vector found in response")
        }
    
        return openAIResponse.Data[0].Embedding, nil
    }
    // Make sure to change CB_USERNAME, CB_PASSWORD, and CB_HOSTNAME to the username, password, and hostname for your database.
    func main() {
        connstr := "couchbases://" + os.Getenv("CB_HOSTNAME") 
        username := os.Getenv("CB_USERNAME")
        password := os.Getenv("CB_PASSWORD")
    	// Make sure to change the bucket, and scope names to match where you stored the sample data in your database. 
        bucket_name := "vector-sample"
        scope_name := "color"
    
        cluster, err := gocb.Connect(connstr, gocb.ClusterOptions{
            Authenticator: gocb.PasswordAuthenticator{
                Username: username,
                Password: password,
            },
            SecurityConfig: gocb.SecurityConfig{
                TLSSkipVerify: true, // Disables TLS certificate verification
            },
        })
        if err != nil {
            log.Fatal(err)
        }
    
        bucket := cluster.Bucket(bucket_name)
        err = bucket.WaitUntilReady(5*time.Second, nil)
        if err != nil {
            log.Fatal(err)
        }
    
        scope := bucket.Scope(scope_name)
    	// Change the question to whatever you want to ask.
        question := "What color hides everything like the night?"
        vect, err := generateVector(question)
        if err != nil {
            log.Fatalf("Error generating vector: %v", err)
        }
    
        request := gocb.SearchRequest{
            VectorSearch: vector.NewSearch(
                []*vector.Query{
                vector.NewQuery("embedding_vector_dot", vect),
            },
            nil,
           ),
        }
    	// Change the limit value to return more results. Change the fields array to return different fields from your Search index.
        opts := &gocb.SearchOptions{Limit: 2, Fields: []string{"color","description"}}
    	
    	// Make sure to change the index name to match your Search index. 
        matchResult, err := scope.Search("color-index", request, opts)
        if err != nil {
            log.Fatal(err)
        }
    
        for matchResult.Next() {
            row := matchResult.Row()
            docID := row.ID
            var fields interface{}
            err := row.Fields(&fields)
            if err != nil {
                log.Fatal(err)
            }
            fmt.Printf("Document ID: %s, Fields: %v\n", docID, fields)
        }
    
        if err = matchResult.Err(); err != nil {
            log.Fatal(err)
        }
    }
    import com.couchbase.client.core.error.*;
    import com.couchbase.client.java.*;
    import com.couchbase.client.java.kv.*;
    import com.couchbase.client.java.json.*;
    import com.couchbase.client.java.search.*;
    import com.couchbase.client.java.search.queries.*;
    import com.couchbase.client.java.search.result.*;
    import com.couchbase.client.java.search.vector.*;
    
    import java.io.IOException;
    import java.net.URI;
    import java.net.http.HttpClient;
    import java.net.http.HttpRequest;
    import java.net.http.HttpResponse;
    import java.net.http.HttpResponse.BodyHandlers;
    import java.util.List;
    
    import java.time.Duration;
    import java.util.Map;
    
    public class RunVectorSearchGenerateEmbed {
        // Make sure to replace OPENAI_API_KEY with your own API Key.
        private static final String OPENAI_API_KEY = System.getenv("OPENAI_API_KEY");
        private static final String OPENAI_URL = "https://api.openai.com/v1/embeddings";
    
        public static float[] generateVector(String inputText) {
            HttpClient client = HttpClient.newHttpClient();
    
            JsonObject jsonBody = JsonObject.create()
                    .put("input", inputText)
                    .put("model", "text-embedding-ada-002");
    
            HttpRequest request = HttpRequest.newBuilder()
                    .uri(URI.create(OPENAI_URL))
                    .header("Content-Type", "application/json")
                    .header("Authorization", "Bearer " + OPENAI_API_KEY)
                    .POST(HttpRequest.BodyPublishers.ofString(jsonBody.toString()))
                    .build();
    
            try {
                HttpResponse<String> response = client.send(request, BodyHandlers.ofString());
    
                JsonObject jsonResponse = JsonObject.fromJson(response.body());
                List<Object> embeddingList = jsonResponse.getArray("data")
                    .getObject(0)
                    .getArray("embedding")
                    .toList();
    
                return toFloatArray(embeddingList);
            } catch (IOException | InterruptedException e) {
                e.printStackTrace();
                return null; // or handle more gracefully
            }
        }
    
        private static float[] toFloatArray(List<Object> list) {
            float[] result = new float[list.size()];
            for (int i = 0; i < list.size(); i++) {
                result[i] = ((Number) list.get(i)).floatValue();
            }
            return result;
        }
    
        public static void main(String[] args) {
    // Make sure to change CB_USERNAME, CB_PASSWORD, and CB_HOSTNAME to the username, password, and hostname for your database.
    	String endpoint = "couchbases://" + System.getenv("CB_HOSTNAME") + "?tls_verify=none"; 
    	String username = System.getenv("CB_USERNAME");
    	String password = System.getenv("CB_PASSWORD");
    // Make sure to change the bucket, scope, collection, and index names to match where you stored the sample data in your database. 
    	String bucketName = "vector-sample";
    	String scopeName = "color";
    	String collectionName = "rgb";
    	String searchIndexName = "color-index";
    
    	try {
    	    // Connect to database/cluster with specified credentials
    	    Cluster cluster = Cluster.connect(
    		    endpoint,
    		    ClusterOptions.clusterOptions(username, password).environment(env -> {
    			    // Use the pre-configured profile below to avoid latency issues with your connection.
    			    env.applyProfile("wan-development");
    		    })
    	    );
    
    	    Bucket bucket = cluster.bucket(bucketName);
    	    bucket.waitUntilReady(Duration.ofSeconds(10));
    	    Scope scope = bucket.scope(scopeName);
    	    Collection collection = scope.collection(collectionName);
                // Change the question to whatever you want to ask.
                String question = "What color hides everything like the night?";
                float[] vector = generateVector(question);
    
    	    SearchRequest request = SearchRequest
    	        .create(VectorSearch.create(
                        VectorQuery.create("embedding_vector_dot", vector).numCandidates(2)));
            // Make sure to change the index name to match your Search index. 
    	    SearchResult result = scope.search("color-index", request,
            // Change the limit value to return more results. Change the value or values in fields to return different fields from your Search index.
                    SearchOptions.searchOptions().limit(3).fields("color","description"));
    
    	    for (SearchRow row : result.rows()) {
    	      System.out.println("Found row: " + row);
    	      System.out.println("   Fields: " + row.fieldsAs(Map.class));
    	    } 
    
    	} catch (UnambiguousTimeoutException ex) {
    	    boolean authFailure = ex.toString().contains("Authentication Failure");
    	    if (authFailure) {
    		    System.out.println("Authentication Failure Detected");
    	    } else {
    		    System.out.println("Error:");
    		    System.out.println(ex.getMessage());
    	    }
    	}
        }
    }
    #!/usr/bin/env python3
    
    import os
    import sys
    from couchbase.cluster import Cluster
    from couchbase.options import ClusterOptions
    from couchbase.auth import PasswordAuthenticator
    from couchbase.exceptions import CouchbaseException
    import couchbase.search as search
    from couchbase.options import SearchOptions
    from couchbase.vector_search import VectorQuery, VectorSearch
    from openai import OpenAI
    
    # Change the question to whatever you want to ask
    question = "What color hides everything like the night?"
    
    # Make sure to replace OPENAI_API_KEY with your own API Key
    openai_api_key = os.getenv("OPENAI_API_KEY")
    client = OpenAI()
    
    # Make sure to change CB_USERNAME, CB_PASSWORD, and CB_HOSTNAME to the username, password, and hostname for your database. 
    pa = PasswordAuthenticator(os.getenv("CB_USERNAME"), os.getenv("CB_PASSWORD"))
    cluster = Cluster("couchbases://" + os.getenv("CB_HOSTNAME") + "/?ssl=no_verify", ClusterOptions(pa))
    # Make sure to change the bucket, scope, and index names to match where you stored the sample data in your database. 
    bucket = cluster.bucket("vector-sample")
    scope = bucket.scope("color")
    search_index = "color-index"
    try:
        vector = client.embeddings.create(input = [question], model="text-embedding-ada-002").data[0].embedding
        search_req = search.SearchRequest.create(search.MatchNoneQuery()).with_vector_search(
            VectorSearch.from_vector_query(VectorQuery('embedding_vector_dot', vector, num_candidates=2)))
            # Change the limit value to return more results. Change the fields array to return different fields from your Search index.
        result = scope.search(search_index, search_req, SearchOptions(limit=13,fields=["color", "description"]))
        for row in result.rows():
            print("Found row: {}".format(row))
        print("Reported total rows: {}".format(
            result.metadata().metrics().total_rows()))
    except CouchbaseException as ex:
        import traceback
        traceback.print_exc()

    Next Steps

    You can create a child field or use the Quick Mode editor to update your Vector Search index to include the description field with your search results.

    For example, you could use the following JSON Vector Search index payload to create your Search index. It includes two child field mappings, colorvect_l2 and embedding_vector_dot on two different vector fields in the keyspace’s documents. It also adds 3 normal Search index fields (brightness, color, and description) to add more usable data to the Vector Search index:

    {
      "type": "fulltext-index",
      "name": "vector-sample.color.color-index",
      "sourceType": "gocbcore",
      "sourceName": "vector-sample",
      "planParams": {
        "maxPartitionsPerPIndex": 512,
        "indexPartitions": 1
      },
      "params": {
        "doc_config": {
          "docid_prefix_delim": "",
          "docid_regexp": "",
          "mode": "scope.collection.type_field",
          "type_field": "type"
        },
        "mapping": {
          "analysis": {},
          "default_analyzer": "standard",
          "default_datetime_parser": "dateTimeOptional",
          "default_field": "_all",
          "default_mapping": {
            "dynamic": false,
            "enabled": false
          },
          "default_type": "_default",
          "docvalues_dynamic": false,
          "index_dynamic": false,
          "store_dynamic": false,
          "type_field": "_type",
          "types": {
            "color.rgb": {
              "dynamic": false,
              "enabled": true,
              "properties": {
                "brightness": {
                  "dynamic": false,
                  "enabled": true,
                  "fields": [
                    {
                      "index": true,
                      "name": "brightness",
                      "store": true,
                      "type": "number"
                    }
                  ]
                },
                "color": {
                  "dynamic": false,
                  "enabled": true,
                  "fields": [
                    {
                      "analyzer": "en",
                      "index": true,
                      "name": "color",
                      "store": true,
                      "type": "text"
                    }
                  ]
                },
                "colorvect_dot": {
                  "dynamic": false,
                  "enabled": true,
                  "fields": [
                    {
                      "dims": 3,
                      "index": true,
                      "name": "colorvect_dot",
                      "similarity": "dot_product",
                      "type": "vector"
                    }
                  ]
                },
                "colorvect_l2": {
                  "dynamic": false,
                  "enabled": true,
                  "fields": [
                    {
                      "dims": 3,
                      "index": true,
                      "name": "colorvect_l2",
                      "similarity": "l2_norm",
                      "type": "vector"
                    }
                  ]
                },
                "description": {
                  "dynamic": false,
                  "enabled": true,
                  "fields": [
                    {
                      "analyzer": "en",
                      "index": true,
                      "name": "description",
                      "store": true,
                      "type": "text"
                    }
                  ]
                },
                "embedding_vector_dot": {
                  "dynamic": false,
                  "enabled": true,
                  "fields": [
                    {
                      "dims": 1536,
                      "index": true,
                      "name": "embedding_vector_dot",
                      "similarity": "dot_product",
                      "type": "vector"
                    }
                  ]
                }
              }
            }
          }
        },
        "store": {
          "indexType": "scorch",
          "segmentVersion": 16
        }
      },
      "sourceParams": {}
    }

    Run the example in Example: Semantic Search with Color Descriptions again to see the description paragraphs in your results.

    Vector Search indexes can use the same settings and features as regular Search indexes. If you want to add additional fields and features to your index, see Search Index Features.