Create a collection-DashVector(DashVector)-阿里云帮助中心

This topic describes how to create a collection using the Java SDK.

Prerequisites

A cluster has been created. For more information, see Create a cluster.
An API key has been obtained. For more information, see Manage API keys.
The latest version of the software development kit (SDK) has been installed. For more information, see Install the DashVector SDK.

API definition

Java

// class DashVectorClient

// Create a collection by specifying a name and vector dimensions.
public Response<Void> create(String name, int dimension);

// Create a collection using a CreateCollectionRequest object.
public Response<Void> create(CreateCollectionRequest request);

Examples

Note

To run the code, replace YOUR_API_KEY with your API key and YOUR_CLUSTER_ENDPOINT with your cluster endpoint.

Create a single-vector collection

Java

import com.aliyun.dashvector.DashVectorClient;
import com.aliyun.dashvector.common.DashVectorException;
import com.aliyun.dashvector.models.requests.CreateCollectionRequest;
import com.aliyun.dashvector.models.responses.Response;
import com.aliyun.dashvector.proto.CollectionInfo;
import com.aliyun.dashvector.proto.FieldType;

public class Main {
    public static void main(String[] args) throws DashVectorException {
        DashVectorClient client = new DashVectorClient("YOUR_API_KEY", "YOUR_CLUSTER_ENDPOINT");

        // Create a collection using a CreateCollectionRequest object.
      
        // Create a collection named quickstart with 4 vector dimensions,
        // a vector data type of float (the default),
        // and a distance metric of dotproduct.
        // Predefine multiple fields: name (STRING), weight (FLOAT), age (INT), id (LONG), tags (ARRAY_STRING), numbers (ARRAY_INT), grades (ARRAY_FLOAT), and bankCards (ARRAY_LONG).
        CreateCollectionRequest request = CreateCollectionRequest.builder()
            .name("quickstart")
            .dimension(4)
            .metric(CollectionInfo.Metric.dotproduct)
            .dataType(CollectionInfo.DataType.FLOAT)
            .filedSchema("name", FieldType.STRING)
            .filedSchema("weight", FieldType.FLOAT)
            .filedSchema("age", FieldType.INT)
            .filedSchema("id", FieldType.LONG)
            .filedSchema("tags", FieldType.ARRAY_STRING)
            .filedSchema("numbers", FieldType.ARRAY_INT)
            .filedSchema("grades", FieldType.ARRAY_FLOAT)
            .filedSchema("bankCards", FieldType.ARRAY_LONG)
            .build();
      
        Response<Void> response = client.create(request);

        System.out.println(response);
        // example output:
        // {"code":0,"message":"","requestId":"883716a3-32f8-4220-ae54-245fa9b87bf0"}
    }
}

Create a multi-vector collection

public void createCollection() {
    CreateCollectionRequest createCollectionRequest = CreateCollectionRequest.builder()
      .name("multi_vector_demo")
      .vectors("title", VectorParam.builder().dimension(4).quantizeType("DT_VECTOR_INT8").build())
      .vectors("content", VectorParam.builder().dimension(6).metric(CollectionInfo.Metric.euclidean).build())
      .sparseVectors("abstruct", VectorParam.builder().metric(CollectionInfo.Metric.dotproduct).build())
      .sparseVectors("keywords", VectorParam.builder().metric(CollectionInfo.Metric.dotproduct).build())
      // Sparse vector indexes currently support only the dotproduct metric. You do not need to set the dimension or dtype because the default values are used.
      .filedSchema("author", FieldType.STRING)
      .filedSchema("tags", FieldType.ARRAY_STRING)
      .build();

    Response<Void> createResponse = client.create(createCollectionRequest);
    System.out.println(createResponse);
    assert createResponse.isSuccess();
}

Request parameters

Use CreateCollectionRequestBuilder to construct a CreateCollectionRequest object. The following table describes the available methods.

Method	Required	Default value	Description
name(String name)	Yes	-	The name of the collection to create.
dimension(int dimension)	Yes	-	The vector dimensions.
dataType(CollectionInfo.DataType dataType)	No	DataType.FLOAT	Supported vector data types `DataType.INT` `DataType.FLOAT`
fieldsSchema(Map<String, FieldType> fieldsSchem)	No	-	The field definitions. The following field types are supported: `FieldType.BOOL` `FieldType.STRING` `FieldType.INT` `FieldType.FLOAT` `FieldType.LONG` `FieldType.ARRAY_STRING` `FieldType.ARRAY_INT` `FieldType.ARRAY_FLOAT` `FieldType.ARRAY_LONG`
fieldSchema(String key, FieldType value)	No	-
metric(CollectionInfo.Metric metric)	No	Metric.cosine	Supported distance measures `Metric.cosine` `Metric.euclidean` `Metric.dotproduct` If you set the metric to `cosine`, the `dataType` must be `FLOAT`.
metric(CollectionInfo.Metric metric)	No	Metric.cosine
extraParams(Map<String, String> params)	No	-	Optional parameters: quantize_type: The quantization policy. For more information, see Dynamic vector quantization.
timeout(Integer timeout)	No	-	`timeout == null`: The API call is synchronous. A response is returned after the collection is created. `timeout == -1`: The API call is asynchronous. `timeout >= 0`: The API call is synchronous and waits for a response. If the collection is not created within the specified period, a timeout error is returned.


vectorParam(VectorParam)	No	-	Sets advanced parameters for the vector field, such as enabling quantization. For more information, see VectorParam.
vectors(Map<String, VectorParam>)	No	-	Defines a multi-vector collection. For more information, see Multi-vector search.
sparseVectors(Map<String, VectorParam>)	No	-	Defines a sparse multi-vector collection. For more information, see Multi-vector search.
build()	-	-	Constructs a `CreateCollectionRequest` object.

Note

For more information about the benefits of predefining fields when you create a collection, see Schema-free.
For more information about quantization policies, see Dynamic vector quantization.

Response parameters

Note

The operation returns a Response<Void> object. The Response<Void> object contains information about the result of the operation, as described in the following table.

Method	Type	Description	Example
getCode()	int	The return value. For more information, see Status codes.	0
getMessage()	String	The returned message.	success
getRequestId()	String	The unique ID of the request.	19215409-ea66-4db9-8764-26ce2eb5bb99
isSuccess()	Boolean	Indicates whether the request was successful.	true