Campo dinâmico

Todos os campos definidos no esquema de uma coleção devem ser incluídos nas entidades a inserir. Se pretender que alguns campos sejam opcionais, considere ativar o campo dinâmico. Este tópico descreve como ativar e utilizar o campo dinâmico.

Síntese

No Milvus, é possível criar um esquema de coleção definindo os nomes e tipos de dados para cada campo da coleção. Quando se adiciona um campo ao esquema, é necessário certificar-se de que esse campo está incluído na entidade que se pretende inserir. Se pretender que alguns campos sejam opcionais, uma opção é ativar o campo dinâmico.

O campo dinâmico é um campo reservado chamado $meta, que é do tipo JavaScript Object Notation (JSON). Quaisquer campos nas entidades que não estejam definidos no esquema serão armazenados neste campo JSON reservado como pares de valores chave.

Para uma coleção com o campo dinâmico ativado, pode utilizar chaves no campo dinâmico para filtragem escalar, tal como faria com campos explicitamente definidos no esquema.

Ativar o campo dinâmico

As colecções criadas utilizando o método descrito em Criar coleção instantaneamente têm o campo dinâmico ativado por predefinição. Também é possível ativar o campo dinâmico manualmente ao criar uma coleção com definições personalizadas.

from pymilvus import MilvusClient

client= MilvusClient(uri="http://localhost:19530")

client.create_collection(
    collection_name="my_dynamic_collection",
    dimension=5,
    # highlight-next-line
    enable_dynamic_field=True
)

import io.milvus.v2.client.ConnectConfig;
import io.milvus.v2.client.MilvusClientV2;
import io.milvus.v2.service.collection.request.CreateCollectionReq;

MilvusClientV2 client = new MilvusClientV2(ConnectConfig.builder()
        .uri("http://localhost:19530")
        .build());
        
CreateCollectionReq createCollectionReq = CreateCollectionReq.builder()
    .collectionName("my_dynamic_collection")
    .dimension(5)
    // highlight-next-line
    .enableDynamicField(true)
    .build()
client.createCollection(createCollectionReq);

import { MilvusClient, DataType } from "@zilliz/milvus2-sdk-node";

const client = new Client({
    address: 'http://localhost:19530'
});

await client.createCollection({
    collection_name: "customized_setup_2",
    schema: schema,
    // highlight-next-line
    enable_dynamic_field: true
});

curl --request POST \
--url "${CLUSTER_ENDPOINT}/v2/vectordb/collections/create" \
--header "Authorization: Bearer ${TOKEN}" \
--header "Content-Type: application/json" \
-d '{
    "collectionName": "my_dynamic_collection",
    "dimension": 5,
    "enableDynamicField": true
}'

Utilizar o campo dinâmico

Quando o campo dinâmico está ativado na sua coleção, todos os campos e respectivos valores que não estão definidos no esquema serão armazenados como pares de valores chave no campo dinâmico.

Por exemplo, suponha que o esquema da coleção define apenas dois campos, denominados id e vector, com o campo dinâmico ativado. Agora, insira o seguinte conjunto de dados nesta coleção.

[
    {id: 0, vector: [0.3580376395471989, -0.6023495712049978, 0.18414012509913835, -0.26286205330961354, 0.9029438446296592], color: "pink_8682"},
    {id: 1, vector: [0.19886812562848388, 0.06023560599112088, 0.6976963061752597, 0.2614474506242501, 0.838729485096104], color: "red_7025"},
    {id: 2, vector: [0.43742130801983836, -0.5597502546264526, 0.6457887650909682, 0.7894058910881185, 0.20785793220625592], color: "orange_6781"},
    {id: 3, vector: [0.3172005263489739, 0.9719044792798428, -0.36981146090600725, -0.4860894583077995, 0.95791889146345], color: "pink_9298"},
    {id: 4, vector: [0.4452349528804562, -0.8757026943054742, 0.8220779437047674, 0.46406290649483184, 0.30337481143159106], color: "red_4794"},
    {id: 5, vector: [0.985825131989184, -0.8144651566660419, 0.6299267002202009, 0.1206906911183383, -0.1446277761879955], color: "yellow_4222"},
    {id: 6, vector: [0.8371977790571115, -0.015764369584852833, -0.31062937026679327, -0.562666951622192, -0.8984947637863987], color: "red_9392"},
    {id: 7, vector: [-0.33445148015177995, -0.2567135004164067, 0.8987539745369246, 0.9402995886420709, 0.5378064918413052], color: "grey_8510"},
    {id: 8, vector: [0.39524717779832685, 0.4000257286739164, -0.5890507376891594, -0.8650502298996872, -0.6140360785406336], color: "white_9381"},
    {id: 9, vector: [0.5718280481994695, 0.24070317428066512, -0.3737913482606834, -0.06726932177492717, -0.6980531615588608], color: "purple_4976"}        
]

O conjunto de dados acima contém 10 entidades, cada uma incluindo os campos id, vector, e color. Aqui, o campo color não está definido no esquema. Uma vez que a coleção tem o campo dinâmico ativado, o campo color será armazenado como um par chave-valor no campo dinâmico.

Inserir dados

O código a seguir demonstra como inserir esse conjunto de dados na coleção.

data=[
    {"id": 0, "vector": [0.3580376395471989, -0.6023495712049978, 0.18414012509913835, -0.26286205330961354, 0.9029438446296592], "color": "pink_8682"},
    {"id": 1, "vector": [0.19886812562848388, 0.06023560599112088, 0.6976963061752597, 0.2614474506242501, 0.838729485096104], "color": "red_7025"},
    {"id": 2, "vector": [0.43742130801983836, -0.5597502546264526, 0.6457887650909682, 0.7894058910881185, 0.20785793220625592], "color": "orange_6781"},
    {"id": 3, "vector": [0.3172005263489739, 0.9719044792798428, -0.36981146090600725, -0.4860894583077995, 0.95791889146345], "color": "pink_9298"},
    {"id": 4, "vector": [0.4452349528804562, -0.8757026943054742, 0.8220779437047674, 0.46406290649483184, 0.30337481143159106], "color": "red_4794"},
    {"id": 5, "vector": [0.985825131989184, -0.8144651566660419, 0.6299267002202009, 0.1206906911183383, -0.1446277761879955], "color": "yellow_4222"},
    {"id": 6, "vector": [0.8371977790571115, -0.015764369584852833, -0.31062937026679327, -0.562666951622192, -0.8984947637863987], "color": "red_9392"},
    {"id": 7, "vector": [-0.33445148015177995, -0.2567135004164067, 0.8987539745369246, 0.9402995886420709, 0.5378064918413052], "color": "grey_8510"},
    {"id": 8, "vector": [0.39524717779832685, 0.4000257286739164, -0.5890507376891594, -0.8650502298996872, -0.6140360785406336], "color": "white_9381"},
    {"id": 9, "vector": [0.5718280481994695, 0.24070317428066512, -0.3737913482606834, -0.06726932177492717, -0.6980531615588608], "color": "purple_4976"}
]

res = client.insert(
    collection_name="my_dynamic_collection",
    data=data
)

print(res)

# Output
# {'insert_count': 10, 'ids': [0, 1, 2, 3, 4, 5, 6, 7, 8, 9]}

import com.google.gson.Gson;
import com.google.gson.JsonObject;

import io.milvus.v2.service.vector.request.InsertReq;
import io.milvus.v2.service.vector.response.InsertResp;
   
Gson gson = new Gson();
List<JsonObject> data = Arrays.asList(
        gson.fromJson("{\"id\": 0, \"vector\": [0.3580376395471989, -0.6023495712049978, 0.18414012509913835, -0.26286205330961354, 0.9029438446296592], \"color\": \"pink_8682\"}", JsonObject.class),
        gson.fromJson("{\"id\": 1, \"vector\": [0.19886812562848388, 0.06023560599112088, 0.6976963061752597, 0.2614474506242501, 0.838729485096104], \"color\": \"red_7025\"}", JsonObject.class),
        gson.fromJson("{\"id\": 2, \"vector\": [0.43742130801983836, -0.5597502546264526, 0.6457887650909682, 0.7894058910881185, 0.20785793220625592], \"color\": \"orange_6781\"}", JsonObject.class),
        gson.fromJson("{\"id\": 3, \"vector\": [0.3172005263489739, 0.9719044792798428, -0.36981146090600725, -0.4860894583077995, 0.95791889146345], \"color\": \"pink_9298\"}", JsonObject.class),
        gson.fromJson("{\"id\": 4, \"vector\": [0.4452349528804562, -0.8757026943054742, 0.8220779437047674, 0.46406290649483184, 0.30337481143159106], \"color\": \"red_4794\"}", JsonObject.class),
        gson.fromJson("{\"id\": 5, \"vector\": [0.985825131989184, -0.8144651566660419, 0.6299267002202009, 0.1206906911183383, -0.1446277761879955], \"color\": \"yellow_4222\"}", JsonObject.class),
        gson.fromJson("{\"id\": 6, \"vector\": [0.8371977790571115, -0.015764369584852833, -0.31062937026679327, -0.562666951622192, -0.8984947637863987], \"color\": \"red_9392\"}", JsonObject.class),
        gson.fromJson("{\"id\": 7, \"vector\": [-0.33445148015177995, -0.2567135004164067, 0.8987539745369246, 0.9402995886420709, 0.5378064918413052], \"color\": \"grey_8510\"}", JsonObject.class),
        gson.fromJson("{\"id\": 8, \"vector\": [0.39524717779832685, 0.4000257286739164, -0.5890507376891594, -0.8650502298996872, -0.6140360785406336], \"color\": \"white_9381\"}", JsonObject.class),
        gson.fromJson("{\"id\": 9, \"vector\": [0.5718280481994695, 0.24070317428066512, -0.3737913482606834, -0.06726932177492717, -0.6980531615588608], \"color\": \"purple_4976\"}", JsonObject.class)
);

InsertReq insertReq = InsertReq.builder()
        .collectionName("my_dynamic_collection")
        .data(data)
        .build();

InsertResp insertResp = client.insert(insertReq);
System.out.println(insertResp);

// Output:
//
// InsertResp(InsertCnt=10, primaryKeys=[0, 1, 2, 3, 4, 5, 6, 7, 8, 9])

const { DataType } = require("@zilliz/milvus2-sdk-node")

// 3. Insert some data

var data = [
    {id: 0, vector: [0.3580376395471989, -0.6023495712049978, 0.18414012509913835, -0.26286205330961354, 0.9029438446296592], color: "pink_8682"},
    {id: 1, vector: [0.19886812562848388, 0.06023560599112088, 0.6976963061752597, 0.2614474506242501, 0.838729485096104], color: "red_7025"},
    {id: 2, vector: [0.43742130801983836, -0.5597502546264526, 0.6457887650909682, 0.7894058910881185, 0.20785793220625592], color: "orange_6781"},
    {id: 3, vector: [0.3172005263489739, 0.9719044792798428, -0.36981146090600725, -0.4860894583077995, 0.95791889146345], color: "pink_9298"},
    {id: 4, vector: [0.4452349528804562, -0.8757026943054742, 0.8220779437047674, 0.46406290649483184, 0.30337481143159106], color: "red_4794"},
    {id: 5, vector: [0.985825131989184, -0.8144651566660419, 0.6299267002202009, 0.1206906911183383, -0.1446277761879955], color: "yellow_4222"},
    {id: 6, vector: [0.8371977790571115, -0.015764369584852833, -0.31062937026679327, -0.562666951622192, -0.8984947637863987], color: "red_9392"},
    {id: 7, vector: [-0.33445148015177995, -0.2567135004164067, 0.8987539745369246, 0.9402995886420709, 0.5378064918413052], color: "grey_8510"},
    {id: 8, vector: [0.39524717779832685, 0.4000257286739164, -0.5890507376891594, -0.8650502298996872, -0.6140360785406336], color: "white_9381"},
    {id: 9, vector: [0.5718280481994695, 0.24070317428066512, -0.3737913482606834, -0.06726932177492717, -0.6980531615588608], color: "purple_4976"}        
]

var res = await client.insert({
    collection_name: "quick_setup",
    data: data,
})

console.log(res.insert_cnt)

// Output
// 
// 10
// 

export CLUSTER_ENDPOINT="http://localhost:19530"
export TOKEN="root:Milvus"

curl --request POST \
--url "${CLUSTER_ENDPOINT}/v2/vectordb/entities/insert" \
--header "Authorization: Bearer ${TOKEN}" \
--header "Content-Type: application/json" \
-d '{
    "data": [
        {"id": 0, "vector": [0.3580376395471989, -0.6023495712049978, 0.18414012509913835, -0.26286205330961354, 0.9029438446296592], "color": "pink_8682"},
        {"id": 1, "vector": [0.19886812562848388, 0.06023560599112088, 0.6976963061752597, 0.2614474506242501, 0.838729485096104], "color": "red_7025"},
        {"id": 2, "vector": [0.43742130801983836, -0.5597502546264526, 0.6457887650909682, 0.7894058910881185, 0.20785793220625592], "color": "orange_6781"},
        {"id": 3, "vector": [0.3172005263489739, 0.9719044792798428, -0.36981146090600725, -0.4860894583077995, 0.95791889146345], "color": "pink_9298"},
        {"id": 4, "vector": [0.4452349528804562, -0.8757026943054742, 0.8220779437047674, 0.46406290649483184, 0.30337481143159106], "color": "red_4794"},
        {"id": 5, "vector": [0.985825131989184, -0.8144651566660419, 0.6299267002202009, 0.1206906911183383, -0.1446277761879955], "color": "yellow_4222"},
        {"id": 6, "vector": [0.8371977790571115, -0.015764369584852833, -0.31062937026679327, -0.562666951622192, -0.8984947637863987], "color": "red_9392"},
        {"id": 7, "vector": [-0.33445148015177995, -0.2567135004164067, 0.8987539745369246, 0.9402995886420709, 0.5378064918413052], "color": "grey_8510"},
        {"id": 8, "vector": [0.39524717779832685, 0.4000257286739164, -0.5890507376891594, -0.8650502298996872, -0.6140360785406336], "color": "white_9381"},
        {"id": 9, "vector": [0.5718280481994695, 0.24070317428066512, -0.3737913482606834, -0.06726932177492717, -0.6980531615588608], "color": "purple_4976"}        
    ],
    "collectionName": "my_dynamic_collection"
}'

# {
#     "code": 0,
#     "data": {
#         "insertCount": 10,
#         "insertIds": [
#             0,
#             1,
#             2,
#             3,
#             4,
#             5,
#             6,
#             7,
#             8,
#             9
#         ]
#     }
# }

Consulta e pesquisa com campo dinâmico

O Milvus suporta a utilização de expressões de filtragem durante as consultas e pesquisas, permitindo-lhe especificar quais os campos a incluir nos resultados. O exemplo seguinte demonstra como efetuar consultas e pesquisas com o campo color, que não está definido no esquema, utilizando o campo dinâmico.

query_vector = [0.3580376395471989, -0.6023495712049978, 0.18414012509913835, -0.26286205330961354, 0.9029438446296592]

res = client.search(
    collection_name="my_dynamic_collection",
    data=[query_vector],
    limit=5,
    # highlight-start
    filter='color like "red%"',
    output_fields=["color"]
    # highlight-end
)

print(res)

# Output
# data: ["[{'id': 1, 'distance': 0.6290165185928345, 'entity': {'color': 'red_7025'}}, {'id': 4, 'distance': 0.5975797176361084, 'entity': {'color': 'red_4794'}}, {'id': 6, 'distance': -0.24996188282966614, 'entity': {'color': 'red_9392'}}]"] 

import io.milvus.v2.service.vector.request.SearchReq
import io.milvus.v2.service.vector.request.data.FloatVec;
import io.milvus.v2.service.vector.response.SearchResp

FloatVec queryVector = new FloatVec(new float[]{0.3580376395471989f, -0.6023495712049978f, 0.18414012509913835f, -0.26286205330961354f, 0.9029438446296592f});
SearchResp resp = client.search(SearchReq.builder()
        .collectionName("my_dynamic_collection")
        .annsField("vector")
        .data(Collections.singletonList(queryVector))
        .outputFields(Collections.singletonList("color"))
        .filter("color like \"red%\"")
        .topK(5)
        .consistencyLevel(ConsistencyLevel.STRONG)
        .build());

System.out.println(resp.getSearchResults());

// Output
//
// [[SearchResp.SearchResult(entity={color=red_7025}, score=0.6290165, id=1), SearchResp.SearchResult(entity={color=red_4794}, score=0.5975797, id=4), SearchResp.SearchResult(entity={color=red_9392}, score=-0.24996188, id=6)]]

const query_vector = [0.3580376395471989, -0.6023495712049978, 0.18414012509913835, -0.26286205330961354, 0.9029438446296592]

res = await client.search({
    collection_name: "quick_setup",
    data: [query_vector],
    limit: 5,
    // highlight-start
    filters: "color like \"red%\"",
    output_fields: ["color"]
    // highlight-end
})

export CLUSTER_ENDPOINT="http://localhost:19530"
export TOKEN="root:Milvus"

curl --request POST \
--url "${CLUSTER_ENDPOINT}/v2/vectordb/entities/search" \
--header "Authorization: Bearer ${TOKEN}" \
--header "Content-Type: application/json" \
-d '{
    "collectionName": "my_dynamic_collection",
    "data": [
        [0.3580376395471989, -0.6023495712049978, 0.18414012509913835, -0.26286205330961354, 0.9029438446296592]
    ],
    "annsField": "vector",
    "filter": "color like \"red%\"",
    "limit": 3,
    "outputFields": ["color"]
}'
# {"code":0,"cost":0,"data":[{"color":"red_7025","distance":0.6290165,"id":1},{"color":"red_4794","distance":0.5975797,"id":4},{"color":"red_9392","distance":-0.24996185,"id":6}]}

Na expressão de filtro utilizada no exemplo de código acima, color like "red%" and likes > 50, as condições especificam que o valor do campo color deve começar por "red". Nos dados de amostra, apenas duas entidades atendem a essa condição. Assim, quando limit (topK) é definido como 3 ou menos, ambas as entidades serão retornadas.

[
    {
        "id": 4, 
        "distance": 0.3345786594834839,
        "entity": {
            "vector": [0.4452349528804562, -0.8757026943054742, 0.8220779437047674, 0.46406290649483184, 0.30337481143159106], 
            "color": "red_4794", 
            "likes": 122
        }
    },
    {
        "id": 6, 
        "distance": 0.6638239834383389，
        "entity": {
            "vector": [0.8371977790571115, -0.015764369584852833, -0.31062937026679327, -0.562666951622192, -0.8984947637863987], 
            "color": "red_9392", 
            "likes": 58
        }
    },
]