DashVector:Getting Started

Core example

The following Python snippet shows the minimal end-to-end flow: create a collection, encode a document, insert it, and run a hybrid query.

import dashvector
from dashtext import SparseVectorEncoder
from dashvector import Doc

# Connect to DashVector
client = dashvector.Client(
    api_key='YOUR_API_KEY',
    endpoint='YOUR_CLUSTER_ENDPOINT'
)

# Create a collection with dot product metric (required for sparse vectors)
client.create('hybrid_collection', dimension=4, metric='dotproduct')
collection = client.get('hybrid_collection')

# Build a sparse vector encoder and encode a document
encoder = SparseVectorEncoder.default()
doc_sparse = encoder.encode_documents("Your document text here.")

# Insert the document with both dense and sparse vectors
collection.insert(Doc(
    id='doc1',
    vector=[0.1, 0.2, 0.3, 0.4],
    sparse_vector=doc_sparse
))

# Search with both dense and sparse vectors
query_sparse = encoder.encode_queries("Your search query here.")
results = collection.query(
    vector=[0.1, 0.1, 0.1, 0.1],
    sparse_vector=query_sparse
)

The sections below walk through each step in detail, with both Python and Java examples.

Prerequisites

Before you begin, make sure you have:

• A DashVector cluster with an available API key and endpoint. For more information, see Vector introduction

• Python or Java SDK installed (dashvector and dashtext packages for Python)

Step 1: Create a collection that supports sparse vectors

Sparse vectors require the dot product distance metric. Create a collection with metric='dotproduct' to enable sparse vector support.

Python

import dashvector

client = dashvector.Client(
    api_key='YOUR_API_KEY',
    endpoint='YOUR_CLUSTER_ENDPOINT'
)
assert client

# Create a collection with dot product metric (required for sparse vectors)
ret = client.create('hybrid_collection', dimension=4, metric='dotproduct')
assert ret

collection = client.get('hybrid_collection')
assert collection

Java

import com.aliyun.dashvector.DashVectorClient;
import com.aliyun.dashvector.DashVectorCollection;
import com.aliyun.dashvector.models.requests.CreateCollectionRequest;
import com.aliyun.dashvector.models.responses.Response;
import com.aliyun.dashvector.proto.CollectionInfo;

DashVectorClient client =
  new DashVectorClient("YOUR_API_KEY", "YOUR_CLUSTER_ENDPOINT");

CreateCollectionRequest request = CreateCollectionRequest.builder()
            .name("hybrid_collection")
            .dimension(4)
            .metric(CollectionInfo.Metric.dotproduct)
            .dataType(CollectionInfo.DataType.FLOAT)
            .build();

Response<Void> response = client.create(request);
System.out.println(response);

DashVectorCollection collection = client.get("hybrid_collection");

Replace YOUR_API_KEY with your API key and YOUR_CLUSTER_ENDPOINT with the endpoint of your cluster for the code to run properly.

Step 2: Build a sparse vector encoder

DashText provides two approaches for creating a SparseVectorEncoder: a built-in encoder for general-purpose text, and a custom encoder trained on your own corpus for higher accuracy with domain-specific terms.

from dashtext import SparseVectorEncoder

encoder = SparseVectorEncoder.default()

import com.aliyun.dashtext.encoder.SparseVectorEncoder;

SparseVectorEncoder encoder = SparseVectorEncoder.getDefaultInstance();

from dashtext import SparseVectorEncoder

encoder = SparseVectorEncoder()

# Your own corpus
corpus = [
    "向量检索服务DashVector基于阿里云自研的高效向量引擎Proxima内核，提供具备水平拓展能力的云原生、全托管的向量检索服务",
    "DashVector将其强大的向量管理、向量查询等多样化能力，通过简洁易用的SDK/API接口透出，方便被上层AI应用迅速集成",
    "从而为包括大模型生态、多模态AI搜索、分子结构分析在内的多种应用场景，提供所需的高效向量检索能力",
    "简单灵活、开箱即用的SDK，使用极简代码即可实现向量管理",
    "自研向量相似性比对算法，快速高效稳定服务",
    "Schema-free设计，通过Schema实现任意条件下的组合过滤查询"
]

# Train the encoder on the full corpus
encoder.train(corpus)

import com.aliyun.dashtext.encoder.SparseVectorEncoder;
import java.util.*;

SparseVectorEncoder encoder = new SparseVectorEncoder();

// Your own corpus
List<String> corpus = Arrays.asList(
  "向量检索服务DashVector基于阿里云自研的高效向量引擎Proxima内核，提供具备水平拓展能力的云原生、全托管的向量检索服务",
  "DashVector将其强大的向量管理、向量查询等多样化能力，通过简洁易用的SDK/API接口透出，方便被上层AI应用迅速集成",
  "从而为包括大模型生态、多模态AI搜索、分子结构分析在内的多种应用场景，提供所需的高效向量检索能力",
  "简单灵活、开箱即用的SDK，使用极简代码即可实现向量管理",
  "自研向量相似性比对算法，快速高效稳定服务",
  "Schema-free设计，通过Schema实现任意条件下的组合过滤查询"
);

// Train the encoder on the full corpus
encoder.train(corpus);

Step 3: Encode and insert a document

Encode document text into a sparse vector with encode_documents, then insert it alongside a dense vector into the collection.

Python

from dashvector import Doc

document = "向量检索服务DashVector基于阿里云自研的高效向量引擎Proxima内核，提供具备水平拓展能力的云原生、全托管的向量检索服务。"
doc_sparse_vector = encoder.encode_documents(document)

print(doc_sparse_vector)
# Output based on the built-in encoder:
# {380823393: 0.7262431704356519, 414191989: 0.7262431704356519, 565176162: 0.7262431704356519, 904594806: 0.7262431704356519, 1005505802: 0.7262431704356519, 1169440797: 0.8883757984694465, 1240922502: 0.7262431704356519, 1313971048: 0.7262431704356519, 1317077351: 0.7262431704356519, 1490140460: 0.7262431704356519, 1574737055: 0.7262431704356519, 1760434515: 0.7262431704356519, 2045788977: 0.8414146776926797, 2141666983: 0.7262431704356519, 2509543087: 0.7262431704356519, 3180265193: 0.7262431704356519, 3845702398: 0.7262431704356519, 4106887295: 0.7262431704356519}

collection.insert(Doc(
    id='A',
    vector=[0.1, 0.2, 0.3, 0.4],
    sparse_vector=doc_sparse_vector
))

Java

String document = "向量检索服务DashVector基于达摩院自研的高效向量引擎Proxima内核，提供具备水平拓展能力的云原生、全托管的向量检索服务。";
Map<Long, Float> sparseVector = encoder.encodeDocuments(document);

System.out.println(sparseVector);
// Output based on the built-in encoder:
// {380823393: 0.7262431704356519, 414191989: 0.7262431704356519, 565176162: 0.7262431704356519, 904594806: 0.7262431704356519, 1005505802: 0.7262431704356519, 1169440797: 0.8883757984694465, 1240922502: 0.7262431704356519, 1313971048: 0.7262431704356519, 1317077351: 0.7262431704356519, 1490140460: 0.7262431704356519, 1574737055: 0.7262431704356519, 1760434515: 0.7262431704356519, 2045788977: 0.8414146776926797, 2141666983: 0.7262431704356519, 2509543087: 0.7262431704356519, 3180265193: 0.7262431704356519, 3845702398: 0.7262431704356519, 4106887295: 0.7262431704356519}

Vector vector = Vector.builder().value(Arrays.asList(0.1f, 0.2f, 0.3f, 0.4f)).build();

// Build a Doc with both dense and sparse vectors
Doc doc = Doc.builder()
  .id("28")
  .sparseVector(sparseVector)
  .vector(vector)
  .build();

// Insert the document
Response<Void> response = collection.insert(InsertDocRequest.builder().doc(doc).build());

Each sparse vector is a map of token hash IDs to BM25 weights. Higher weights indicate more distinctive terms in the document.

Step 4: Run a keyword-aware semantic search

Encode the search query into a sparse vector with encode_queries, then pass both dense and sparse vectors to query.

Python

query = "什么是向量检索服务？"
sparse_vector = encoder.encode_queries(query)

print(sparse_vector)
# Output based on the built-in encoder:
# {1169440797: 0.2947158712590364, 2045788977: 0.7052841287409635}

docs = collection.query(
    vector=[0.1, 0.1, 0.1, 0.1],
    sparse_vector=sparse_vector
)

Java

String query = "什么是向量检索服务？";

Map<Long, Float> sparseVector = encoder.encodeQueries(query);

System.out.println(sparseVector);
// Output based on the built-in encoder:
// {1169440797: 0.2947158712590364, 2045788977: 0.7052841287409635}

Vector vector = Vector.builder().value(Arrays.asList(0.1f, 0.2f, 0.3f, 0.4f)).build();

// Build a query with both dense and sparse vectors
QueryDocRequest request = QueryDocRequest.builder()
  .vector(vector)
  .sparse_ector(sparseVector)
  .topk(100)
  .includeVector(true)
  .build();

Response<List<Doc>> response = collection.query(request);
System.out.println(response);

Step 5: Tune results with weighted vectors

Control the balance between semantic similarity (dense vectors) and keyword matching (sparse vectors) using the alpha parameter:

Apply combine_dense_and_sparse to scale both vectors before querying:

Python

from dashtext import combine_dense_and_sparse

query = "什么是向量检索服务？"
sparse_vector = encoder.encode_queries(query)

# Set alpha to control the balance: 0.0 = keywords only, 1.0 = semantics only
alpha = 0.7
dense_vector = [0.1, 0.1, 0.1, 0.1]
scaled_dense_vector, scaled_sparse_vector = combine_dense_and_sparse(dense_vector, sparse_vector, alpha)

docs = collection.query(
    vector=scaled_dense_vector,
    sparse_vector=scaled_sparse_vector
)

Java

String query = "什么是向量检索服务？";

Map<Long, Float> sparseVector = encoder.encodeQueries(query);

System.out.println(sparseVector);
// Output based on the built-in encoder:
// {1169440797: 0.2947158712590364, 2045788977: 0.7052841287409635}

Vector denseVector = Vector.builder().value(Arrays.asList(0.1f, 0.2f, 0.3f, 0.4f)).build();

// Scale dense and sparse vectors by the alpha weight factor
float alpha = 0.1f;
sparseVector.forEach((key, value) -> sparseVector.put(key, value * alpha));
denseVector = Vector.builder().value(
            denseVector.getValue().stream().map(number -> number.floatValue() * alpha).collect(Collectors.toList())
    ).build();

// Query with weighted vectors
QueryDocRequest request = QueryDocRequest.builder()
  .vector(denseVector)
  .sparse_ector(sparseVector)
  .topk(100)
  .includeVector(true)
  .build();

Response<List<Doc>> response = collection.query(request);
System.out.println(response);

API reference

For the complete DashText API, see DashText SDK for Python.