ICU

The icu tokenizer is built on the Internationalization Components of Unicode (ICU) open‑source project, which provides key tools for software internationalization. By using ICU's word‑break algorithm, the tokenizer can accurately split text into words across the majority of the world’s languages.

Notes

The icu tokenizer preserves punctuation marks and spaces as separate tokens in the output. For example, "Привет! Как дела?" becomes ["Привет", "!", " ", "Как", " ", "дела", "?"]. To remove these standalone punctuation tokens, use the removepunct filter.

Configuration

To configure an analyzer using the icu tokenizer, set tokenizer to icu in analyzer_params.

Python

python

analyzer_params = {
    "tokenizer": "icu",
}

Java

java

Map<String, Object> analyzerParams = new HashMap<>();
analyzerParams.put("tokenizer", "icu");

Go

go

analyzerParams = map[string]any{"tokenizer": "icu"}

Rust

rust

use milvus::v2::prelude::*;
use serde_json::json;

let analyzer_params = json!({ "tokenizer": "icu" });

C++

c++

nlohmann::json analyzer_params = {
    {"tokenizer", "icu"}
};

NodeJS

javascript

const analyzer_params = {
    tokenizer: "icu",
};

cURL

bash

# restful
analyzerParams='{
  "tokenizer": "icu"
}'

The icu tokenizer can work in conjunction with one or more filters. For example, the following code defines an analyzer that uses the icu tokenizer and remove punct filter:

Python

python

analyzer_params = {
    "tokenizer": "icu",
    "filter": ["removepunct"]
}

Java

java

Map<String, Object> analyzerParams = new HashMap<>();
analyzerParams.put("tokenizer", "icu");
analyzerParams.put("filter", Collections.singletonList("removepunct"));

Go

go

analyzerParams = map[string]any{"tokenizer": "icu", "filter": []string{"removepunct"}}

Rust

rust

use milvus::v2::prelude::*;
use serde_json::json;

let analyzer_params = json!({
    "tokenizer": "icu",
    "filter": ["removepunct"]
});

C++

c++

nlohmann::json analyzer_params = {
    {"tokenizer", "icu"},
    {"filter", {"removepunct"}}
};

NodeJS

javascript

const analyzer_params = {
    tokenizer: "icu",
    filter: ["removepunct"],
};

cURL

bash

# restful
analyzerParams='{
  "tokenizer": "icu",
  "filter": ["removepunct"]
}'

After defining analyzer_params, you can apply them to a VARCHAR field when defining a collection schema. This allows Zilliz Cloud to process the text in that field using the specified analyzer for efficient tokenization and filtering. For details, refer to Example use.

Examples

Before applying the analyzer configuration to your collection schema, verify its behavior using the run_analyzer method.

Analyzer configuration

Python

python

analyzer_params = {
    "tokenizer": "icu",
}

Java

java

Map<String, Object> analyzerParams = new HashMap<>();
analyzerParams.put("tokenizer", "icu");

Go

go

analyzerParams = map[string]any{"tokenizer": "icu"}

Rust

rust

use milvus::v2::prelude::*;
use serde_json::json;

let analyzer_params = json!({ "tokenizer": "icu" });

C++

c++

nlohmann::json analyzer_params = {
    {"tokenizer", "icu"}
};

NodeJS

javascript

const analyzer_params = {
    tokenizer: "icu",
};

cURL

bash

# restful
analyzerParams='{
  "tokenizer": "icu"
}'

Verification using run_analyzer

Python

python

from pymilvus import (
    MilvusClient,
)

client = MilvusClient(
    uri="YOUR_CLUSTER_ENDPOINT",
    token="YOUR_CLUSTER_TOKEN"
)

# Sample text to analyze
sample_text = "Привет! Как дела?"

# Run the icu analyzer with the defined configuration
result = client.run_analyzer(sample_text, analyzer_params)
print("ICU analyzer output:", result)

Java

java

import io.milvus.v2.client.ConnectConfig;
import io.milvus.v2.client.MilvusClientV2;
import io.milvus.v2.service.vector.request.RunAnalyzerReq;
import io.milvus.v2.service.vector.response.RunAnalyzerResp;

ConnectConfig config = ConnectConfig.builder()
        .uri("YOUR_CLUSTER_ENDPOINT")
        .token("YOUR_CLUSTER_TOKEN")
        .build();
MilvusClientV2 client = new MilvusClientV2(config);

List<String> texts = new ArrayList<>();
texts.add("Привет! Как дела?");

RunAnalyzerResp resp = client.runAnalyzer(RunAnalyzerReq.builder()
        .texts(texts)
        .analyzerParams(analyzerParams)
        .build());
List<RunAnalyzerResp.AnalyzerResult> results = resp.getResults();

Go

go

import (
    "context"
    "fmt"

    "github.com/milvus-io/milvus/client/v3/milvusclient"
)

client, err := milvusclient.New(ctx, &milvusclient.ClientConfig{
    Address: "YOUR_CLUSTER_ENDPOINT",
    APIKey:  "YOUR_CLUSTER_TOKEN",
})
if err != nil {
    fmt.Println(err.Error())
    // handle error
}

texts := []string{"Привет! Как дела?"}
option := milvusclient.NewRunAnalyzerOption(texts...).
    WithAnalyzerParams(analyzerParams)

result, err := client.RunAnalyzer(ctx, option)
if err != nil {
    fmt.Println(err.Error())
    // handle error
}

Rust

rust

use milvus::v2::prelude::*;
use serde_json::json;

let analyzer_params = json!({ "tokenizer": "icu" });
let result = client
    .run_analyzer(
        RunAnalyzerRequest::builder()
            .texts(vec!["Привет! Как дела?"])
            .analyzer_params(analyzer_params)
            .build()?,
    )
    .await?;

C++

c++

#include "milvus/MilvusClientV2.h"

auto client = milvus::MilvusClientV2::Create();

milvus::ConnectParam connect_param{"YOUR_CLUSTER_ENDPOINT", "YOUR_CLUSTER_TOKEN"};
auto status = client->Connect(connect_param);
if (!status.IsOk()) {
    std::cout << status.Message() << std::endl;
}

std::string text = "Привет! Как дела?";
auto request = milvus::RunAnalyzerRequest()
                       .AddText(text)
                       .WithAnalyzerParams(analyzer_params);

milvus::RunAnalyzerResponse response;
status = client->RunAnalyzer(request, response);
if (!status.IsOk()) {
    std::cout << status.Message() << std::endl;
}

NodeJS

javascript

const sample_text = "Привет! Как дела?";

// Run the icu analyzer with the defined configuration
const result = await client.runAnalyzer({
    text: sample_text,
    analyzer_params,
});
console.log("ICU analyzer output:", result);

cURL

bash

# restful
curl --request POST \
     --url "YOUR_CLUSTER_ENDPOINT/v2/vectordb/common/run_analyzer" \
     --header "Authorization: Bearer YOUR_CLUSTER_TOKEN" \
     --header "Content-Type: application/json" \
     -d '{
        "text": ["Привет! Как дела?"],
        "analyzerParams": "{\"tokenizer\": \"icu\"}"
     }'

Expected output

plaintext

['Привет', '!', ' ', 'Как', ' ', 'дела', '?']