词干

stemmer 过滤器可将单词还原为其基本形式或词根形式（称为词干化），从而更容易匹配不同词性中含义相似的单词。stemmer 过滤器支持多种语言，可在各种语言环境中进行有效搜索和索引。

配置

stemmer 过滤器是 Milvus 的自定义过滤器。要使用它，请在过滤器配置中指定"type": "stemmer" ，并使用language 参数选择所需的语言进行词干处理。

Python

analyzer_params = {
    "tokenizer": "standard",
    "filter":[{
        "type": "stemmer", # Specifies the filter type as stemmer
        "language": "english", # Sets the language for stemming to English
    }],
}

Java

Map<String, Object> analyzerParams = new HashMap<>();
analyzerParams.put("tokenizer", "standard");
analyzerParams.put("filter",
        Collections.singletonList(
                new HashMap<String, Object>() {{
                    put("type", "stemmer");
                    put("language", "english");
                }}
        )
);

NodeJS

const analyzer_params = {
    "tokenizer": "standard",
    "filter":[{
        "type": "stemmer", // Specifies the filter type as stop
        "language": "english",
    }],
};

Go

analyzerParams = map[string]any{"tokenizer": "standard",
    "filter": []any{map[string]any{
        "type":     "stemmer",
        "language": "english",
    }}}

cURL

# restful
analyzerParams='{
  "tokenizer": "standard",
  "filter": [
    {
      "type": "stemmer",
      "language": "english"
    }
  ]
}'

stemmer 过滤器接受以下可配置参数。

| 参数

|

参数

| | --- | --- | |

language

|

指定词干处理的语言。支持的语言包括"arabic","danish","dutch","english","finnish","french","german","greek","hungarian","italian","norwegian","portuguese","romanian","russian","spanish","swedish","tamil" 、"turkish"

|

stemmer 过滤器对标记符生成的术语进行操作，因此必须与标记符结合使用。

定义analyzer_params 后，可以在定义 Collection Schema 时将它们应用到VARCHAR 字段。这样，Milvus 就可以使用指定的分析器对该字段中的文本进行处理，从而实现高效的标记化和过滤。有关详情，请参阅示例使用。

示例

在将分析器配置应用到 Collection 模式之前，请使用run_analyzer 方法验证其行为。

分析器配置

Python

analyzer_params = {
    "tokenizer": "standard",
    "filter":[{
        "type": "stemmer", # Specifies the filter type as stemmer
        "language": "english", # Sets the language for stemming to English
    }],
}

Java

Map<String, Object> analyzerParams = new HashMap<>();
analyzerParams.put("tokenizer", "standard");
analyzerParams.put("filter",
        Collections.singletonList(
                new HashMap<String, Object>() {{
                    put("type", "stemmer");
                    put("language", "english");
                }}
        )
);

NodeJS

// javascript

Go

analyzerParams = map[string]any{"tokenizer": "standard",
    "filter": []any{map[string]any{
        "type":     "stemmer",
        "language": "english",
    }}}

cURL

# restful
analyzerParams='{
  "tokenizer": "standard",
  "filter": [
    {
      "type": "stemmer",
      "language": "english"
    }
  ]
}'

验证使用run_analyzerCompatible with Milvus 2.5.11+

Python

from pymilvus import (
    MilvusClient,
)

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

# Sample text to analyze
sample_text = "running runs looked ran runner"

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

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

List<String> texts = new ArrayList<>();
texts.add("running runs looked ran runner");

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

NodeJS

// javascript

Go

import (
    "context"
    "encoding/json"
    "fmt"

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

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

bs, _ := json.Marshal(analyzerParams)
texts := []string{"running runs looked ran runner"}
option := milvusclient.NewRunAnalyzerOption(texts).
    WithAnalyzerParams(string(bs))

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

# restful
not support yet

预期输出

['run', 'run', 'look', 'ran', 'runner']