Standard Tokenizer

The standard tokenizer in Milvus groups consecutive Unicode letters and numeric characters into tokens, splitting at other characters.

Tokenization rules

The standard tokenizer keeps consecutive characters that belong to the following sets in the same token:

• ASCII characters: letters A-Z and a-z, and digits 0-9.
• Non-ASCII characters: characters with the Unicode Alphabetic property or one of the numeric general categories Nd, Nl, or No.

Unicode property or category	Meaning	Examples of characters retained in tokens
Alphabetic	Letters across writing systems, including Chinese characters and Japanese kana, and some combining marks	中文测试, カタカナ
Nd (Decimal_Number)	Decimal digits	٣
Nl (Letter_Number)	Letter-like numeric characters	Ⅷ
No (Other_Number)	Other numeric characters, such as circled numbers, superscripts, and fractions	①²¾

Characters outside these sets separate tokens and are discarded. These include whitespace, punctuation, underscores (_), hyphens (-), apostrophes ('), and symbols such as +, $, and 😀. Consecutive separators do not produce empty tokens.

Character classification follows Rust’s char::is_alphanumeric(). For property definitions, see Unicode Standard Annex #44. The complete Unicode 17.0 character lists are available in DerivedCoreProperties.txt for Alphabetic and DerivedGeneralCategory.txt for Nd, Nl, and No. Character membership depends on the Unicode data used by the deployed version.

The following examples use {"tokenizer": "standard"} with no filters. The tokenizer preserves letter case and does not segment continuous Chinese text into individual words.

Input	Output tokens
foo_bar-can't😀123	["foo", "bar", "can", "t", "123"]
中文测试	["中文测试"]
version①.¾	["version①", "¾"]
Hello,World!	["Hello", "World"]

Configuration

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

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

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

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

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

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

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

analyzer_params = {
    "tokenizer": "standard",
    "filter": ["lowercase"]
}

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

const analyzer_params = {
    "tokenizer": "standard",
    "filter": ["lowercase"]
};

analyzerParams = map[string]any{"tokenizer": "standard", "filter": []any{"lowercase"}}

# restful
analyzerParams='{
  "tokenizer": "standard",
  "filter": [
    "lowercase"
  ]
}'

After defining analyzer_params, you can apply them to a VARCHAR field when defining a collection schema. This allows Milvus 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

analyzer_params = {
    "tokenizer": "standard",
    "filter": ["lowercase"]
}

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

// javascript

analyzerParams = map[string]any{"tokenizer": "standard", "filter": []any{"lowercase"}}

# restful

Verification using run_analyzer

from pymilvus import (
    MilvusClient,
)

client = MilvusClient(
    uri="http://localhost:19530",
    token="root:Milvus"
)

# Sample text to analyze
sample_text = "The Milvus vector database is built for scale!"

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

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("http://localhost:19530")
        .token("root:Milvus")
        .build();
MilvusClientV2 client = new MilvusClientV2(config);

List<String> texts = new ArrayList<>();
texts.add("The Milvus vector database is built for scale!");

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

// javascript

import (
    "context"
    "fmt"

    "github.com/milvus-io/milvus/client/v2/milvusclient"
)

ctx := context.Background()
client, err := milvusclient.New(ctx, &milvusclient.ClientConfig{
    Address: "localhost:19530",
    APIKey:  "root:Milvus",
})
if err != nil {
    fmt.Println(err.Error())
    // handle error
}

texts := []string{"The Milvus vector database is built for scale!"}
option := milvusclient.NewRunAnalyzerOption(texts...).
    WithAnalyzerParams(analyzerParams)

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

# restful

Expected output

['the', 'milvus', 'vector', 'database', 'is', 'built', 'for', 'scale']