018.Elasticsearch分词器原理及使用

CoderJed

发布于 2020-07-06 15:47:53

1.4K0

发布于 2020-07-06 15:47:53

文章被收录于专栏：Jed的技术阶梯

1.分词器介绍

什么是分词器？将一段文本按照一定的逻辑，分析成多个词语，同时对这些词语进行常规化(normalization)的一种工具，例如： "hello tom and jerry"可以分为"hello"、"tom"、"and"、"jerry"这4个单词常规化是说，例如，"hello tom & jerry"，那么把"&"这个字符转换为"and"，对一个html标签进行分词时，先去掉标签"<span>hello<span>" -> "hello"
常用的内置分词器
- standard analyzer
- simple analyzer
- whitespace analyzer
- stop analyzer
- language analyzer
- pattern analyzer

1.1 standard analyzer

默认分词器：按照非字母和非数字字符进行分隔，单词转为小写测试文本：a*B!c d4e 5f 7-h 分词结果：a、b、c、d4e、5f、7、h

{
  "tokens" : [
    {
      "token" : "a", # 分词后的单词
      "start_offset" : 0, # 在原文本中的起始位置
      "end_offset" : 1, # 原文本中的结束位置
      "type" : "<ALPHANUM>", # 单词类型：ALPHANUM(字母)、NUM(数字)
      "position" : 0 # 单词位置,是分出来的所有单词的第几个单词
    },
    {
      "token" : "b",
      "start_offset" : 2,
      "end_offset" : 3,
      "type" : "<ALPHANUM>",
      "position" : 1
    },
    {
      "token" : "c",
      "start_offset" : 4,
      "end_offset" : 5,
      "type" : "<ALPHANUM>",
      "position" : 2
    },
    {
      "token" : "d4e",
      "start_offset" : 6,
      "end_offset" : 9,
      "type" : "<ALPHANUM>",
      "position" : 3
    },
    {
      "token" : "5f",
      "start_offset" : 10,
      "end_offset" : 12,
      "type" : "<ALPHANUM>",
      "position" : 4
    },
    {
      "token" : "7",
      "start_offset" : 13,
      "end_offset" : 14,
      "type" : "<NUM>",
      "position" : 5
    },
    {
      "token" : "h",
      "start_offset" : 15,
      "end_offset" : 16,
      "type" : "<ALPHANUM>",
      "position" : 6
    }
  ]
}

1.2 simple analyzer

分词效果：按照非字母字符进行分隔，单词转为小写测试文本：a*B!c d4e 5f 7-h 分词结果：a、b、c、d、e、f、h

GET _analyze
{
  "analyzer": "simple",
  "text": "a*B!c d4e 5f 7-h"
}

{
  "tokens" : [
    {
      "token" : "a",
      "start_offset" : 0,
      "end_offset" : 1,
      "type" : "word",
      "position" : 0
    },
    {
      "token" : "b",
      "start_offset" : 2,
      "end_offset" : 3,
      "type" : "word",
      "position" : 1
    },
    {
      "token" : "c",
      "start_offset" : 4,
      "end_offset" : 5,
      "type" : "word",
      "position" : 2
    },
    {
      "token" : "d",
      "start_offset" : 6,
      "end_offset" : 7,
      "type" : "word",
      "position" : 3
    },
    {
      "token" : "e",
      "start_offset" : 8,
      "end_offset" : 9,
      "type" : "word",
      "position" : 4
    },
    {
      "token" : "f",
      "start_offset" : 11,
      "end_offset" : 12,
      "type" : "word",
      "position" : 5
    },
    {
      "token" : "h",
      "start_offset" : 15,
      "end_offset" : 16,
    "type" : "word",
      "position" : 6
    }
  ]
}

1.3 whitespace analyzer

分词效果：按照空白字符进行分隔测试文本：a*B!c D d4e 5f 7-h 分词结果：a*B!c、D、d4e、5f、7-h

GET _analyze
{
  "analyzer": "whitespace",
  "text": "a*B!c D d4e 5f 7-h"
}

{
  "tokens" : [
    {
      "token" : "a*B!c",
      "start_offset" : 0,
      "end_offset" : 5,
      "type" : "word",
      "position" : 0
    },
    {
      "token" : "D",
      "start_offset" : 6,
      "end_offset" : 7,
      "type" : "word",
      "position" : 1
    },
    {
      "token" : "d4e",
      "start_offset" : 8,
      "end_offset" : 11,
      "type" : "word",
      "position" : 2
    },
    {
      "token" : "5f",
      "start_offset" : 12,
      "end_offset" : 14,
      "type" : "word",
      "position" : 3
    },
    {
      "token" : "7-h",
      "start_offset" : 15,
      "end_offset" : 18,
      "type" : "word",
      "position" : 4
    }
  ]
}

1.4 stop analyzer

分词效果：使用非字母字符进行分隔，单词转换为小写，并去掉停用词(默认为英语的停用词，例如the、a、an、this、of、at等) 测试文本：The apple is red 分词结果：apple、red

GET _analyze
{
  "analyzer": "stop",
  "text": "The apple is red"
}

{
  "tokens" : [
    {
      "token" : "apple",
      "start_offset" : 4,
      "end_offset" : 9,
      "type" : "word",
      "position" : 1
    },
    {
      "token" : "red",
      "start_offset" : 13,
      "end_offset" : 16,
      "type" : "word",
      "position" : 3
    }
  ]
}

1.5 language analyzer

分词效果：使用指定的语言的语法进行分词，默认为english，没有内置中文分词器

GET _analyze
{
  "analyzer": "english",
  "text": "\"I'm Tony,\", he said, \"nice to meet you!\""
}

{
  "tokens" : [
    {
      "token" : "i'm",
      "start_offset" : 1,
      "end_offset" : 4,
      "type" : "<ALPHANUM>",
      "position" : 0
    },
    {
      "token" : "toni",
      "start_offset" : 5,
      "end_offset" : 9,
      "type" : "<ALPHANUM>",
      "position" : 1
    },
    {
      "token" : "he",
      "start_offset" : 13,
      "end_offset" : 15,
      "type" : "<ALPHANUM>",
      "position" : 2
    },
    {
      "token" : "said",
      "start_offset" : 16,
      "end_offset" : 20,
      "type" : "<ALPHANUM>",
      "position" : 3
    },
    {
      "token" : "nice",
      "start_offset" : 23,
      "end_offset" : 27,
      "type" : "<ALPHANUM>",
      "position" : 4
    },
    {
      "token" : "meet",
      "start_offset" : 31,
      "end_offset" : 35,
      "type" : "<ALPHANUM>",
      "position" : 6
    },
    {
      "token" : "you",
      "start_offset" : 36,
      "end_offset" : 39,
      "type" : "<ALPHANUM>",
      "position" : 7
    }
  ]
}

1.6 pattern analyzer

分词效果：使用指定的正则表达式进行分词，默认\\W+，即多个非数字非字母字符

GET _analyze
{
  "analyzer": "pattern",
  "text": "The best 3-points shooter is Curry!"
}

{
  "tokens" : [
    {
      "token" : "the",
      "start_offset" : 0,
      "end_offset" : 3,
      "type" : "word",
      "position" : 0
    },
    {
      "token" : "best",
      "start_offset" : 4,
      "end_offset" : 8,
      "type" : "word",
      "position" : 1
    },
    {
      "token" : "3",
      "start_offset" : 9,
      "end_offset" : 10,
      "type" : "word",
      "position" : 2
    },
    {
      "token" : "points",
      "start_offset" : 11,
      "end_offset" : 17,
      "type" : "word",
      "position" : 3
    },
    {
      "token" : "shooter",
      "start_offset" : 18,
      "end_offset" : 25,
      "type" : "word",
      "position" : 4
    },
    {
      "token" : "is",
      "start_offset" : 26,
      "end_offset" : 28,
      "type" : "word",
      "position" : 5
    },
    {
      "token" : "curry",
      "start_offset" : 29,
      "end_offset" : 34,
      "type" : "word",
      "position" : 6
    }
  ]
}

2.分词器使用

2.1 指定index的分词器

创建测试索引

PUT /my_index
{
  "settings": {
    "analysis": {
      "analyzer": {
        "my_analyzer_1": {
          "type": "whitespace"
        }
      }
    }
  },
  "mappings": {
    "_doc": {
      "properties": {
        "id": {
          "type": "keyword"
        },
        "name": {
          "type": "text"
        },
        "desc": {
          "type": "text",
          "analyzer": "my_analyzer_1"
        }
      }
    }
  }
}

创建测试数据：

PUT my_index/_doc/1
{
  "id": "001",
  "name": "Curry",
  "desc": "The best 3-points shooter is Curry!"
}

查询：由于desc字段使用whitespace分词，所以通过curry是查询不到的，需要通过Curry!来查询

GET my_index/_search
{
  "query": {
    "match": {
      "desc": "curry"
    }
  }
}

{
  "took" : 2,
  "timed_out" : false,
  "_shards" : {
    "total" : 5,
    "successful" : 5,
    "skipped" : 0,
    "failed" : 0
  },
  "hits" : {
    "total" : 0,
    "max_score" : null,
    "hits" : [ ]
  }
}

GET my_index/_search
{
  "query": {
    "match": {
      "desc": "Curry!"
    }
  }
}

{
  "took" : 4,
  "timed_out" : false,
  "_shards" : {
    "total" : 5,
    "successful" : 5,
    "skipped" : 0,
    "failed" : 0
  },
  "hits" : {
    "total" : 1,
    "max_score" : 0.2876821,
    "hits" : [
      {
        "_index" : "my_index",
        "_type" : "_doc",
        "_id" : "1",
        "_score" : 0.2876821,
        "_source" : {
          "id" : "001",
          "name" : "Curry",
          "desc" : "The best 3-points shooter is Curry!"
        }
      }
    ]
  }
}

2.2 更改分词器设置

# 创建索引，并设置分词器，启用停用词，默认的standard分词器是没有使用停用词的
PUT /my_index
{
  "settings": {
    "analysis": {
      "analyzer": {
        "my_standard": {
          "type": "standard",
          "stopwords": "_english_"
        }
      }
    }
  }
}

# 测试
GET /my_index/_analyze
{
  "analyzer": "my_standard",
  "text": "a dog is in the house"
}

{
  "tokens": [
    {
      "token": "dog",
      "start_offset": 2,
      "end_offset": 5,
      "type": "<ALPHANUM>",
      "position": 1
    },
    {
      "token": "house",
      "start_offset": 16,
      "end_offset": 21,
      "type": "<ALPHANUM>",
      "position": 5
    }
  ]
}

2.3 自定义分词器

PUT /my_index
{
  "settings": {
    "analysis": {
      "char_filter": {
        "my_char_filter": {
          "type": "mapping",
          "mappings": ["& => and"] # "$"转换为"and"
        }
      },
      "filter": {
        "my_filter": {
          "type": "stop",
          "stopwords": ["the", "a"] # 指定两个停用词
        }
      },
      "analyzer": {
        "my_analyzer": {
          "type": "custom",
          "char_filter": ["html_strip", "my_char_filter"], # 使用内置的html标签过滤和自定义的my_char_filter
          "tokenizer": "standard",
          "filter": ["lowercase", "my_filter"] # 使用内置的lowercase filter和自定义的my_filter
        }
      }
    }
  }
}

GET /my_index/_analyze
{
  "analyzer": "my_analyzer",
  "text": "tom&jerry are a friend in the house, <a>, HAHA!!"
｝

{
  "tokens": [
    {
      "token": "tomandjerry",
      "start_offset": 0,
      "end_offset": 9,
      "type": "<ALPHANUM>",
      "position": 0
    },
    {
      "token": "are",
      "start_offset": 10,
      "end_offset": 13,
      "type": "<ALPHANUM>",
      "position": 1
    },
    {
      "token": "friend",
      "start_offset": 16,
      "end_offset": 22,
      "type": "<ALPHANUM>",
      "position": 3
    },
    {
      "token": "in",
      "start_offset": 23,
      "end_offset": 25,
      "type": "<ALPHANUM>",
      "position": 4
    },
    {
      "token": "house",
      "start_offset": 30,
      "end_offset": 35,
      "type": "<ALPHANUM>",
      "position": 6
    },
    {
      "token": "haha",
      "start_offset": 42,
      "end_offset": 46,
      "type": "<ALPHANUM>",
      "position": 7
    }
  ]
}

2.4 为指定的type、指定的字段设置自定义的分词器

PUT /my_index/_mapping/my_type
{
  "properties": {
    "content": {
      "type": "text",
      "analyzer": "my_analyzer"
    }
  }
}

3. 中文分词器

3.1. 中文分词器介绍

Elasticsearch内置的分词器无法对中文进行分词，例如：

GET _analyze
{
  "analyzer": "standard",
  "text": "火箭明年总冠军"
}

{
  "tokens" : [
    {
      "token" : "火",
      "start_offset" : 0,
      "end_offset" : 1,
      "type" : "<IDEOGRAPHIC>",
      "position" : 0
    },
    {
      "token" : "箭",
      "start_offset" : 1,
      "end_offset" : 2,
      "type" : "<IDEOGRAPHIC>",
      "position" : 1
    },
    {
      "token" : "明",
      "start_offset" : 2,
      "end_offset" : 3,
      "type" : "<IDEOGRAPHIC>",
      "position" : 2
    },
    {
      "token" : "年",
      "start_offset" : 3,
      "end_offset" : 4,
      "type" : "<IDEOGRAPHIC>",
      "position" : 3
    },
    {
      "token" : "总",
      "start_offset" : 4,
      "end_offset" : 5,
      "type" : "<IDEOGRAPHIC>",
      "position" : 4
    },
    {
      "token" : "冠",
      "start_offset" : 5,
      "end_offset" : 6,
      "type" : "<IDEOGRAPHIC>",
      "position" : 5
    },
    {
      "token" : "军",
      "start_offset" : 6,
      "end_offset" : 7,
      "type" : "<IDEOGRAPHIC>",
      "position" : 6
    }
  ]
}

我们期望的分词结果是火箭、明年、总冠军，这就需要使用中文分词器了。

常见的中文分词器
- smartCN ：一个简单的中⽂或中英⽂混合文本分词器
- IK分词器：更智能更友好的中⽂分词器

3.2 smartCN安装方式

bin/elasticsearch-plugin install analysis-smartcn

完成后重启ES集群，测试：

GET _analyze
{
  "analyzer": "smartcn",
  "text": "火箭明年总冠军"
}

{
  "tokens" : [
    {
      "token" : "火箭",
      "start_offset" : 0,
      "end_offset" : 2,
      "type" : "word",
      "position" : 0
    },
    {
      "token" : "明年",
      "start_offset" : 2,
      "end_offset" : 4,
      "type" : "word",
      "position" : 1
    },
    {
      "token" : "总",
      "start_offset" : 4,
      "end_offset" : 5,
      "type" : "word",
      "position" : 2
    },
    {
      "token" : "冠军",
      "start_offset" : 5,
      "end_offset" : 7,
      "type" : "word",
      "position" : 3
    }
  ]
}

3.3 IK分词器安装

下载地址：https://github.com/medcl/elasticsearch-analysis-ik/releases

下载与ES同版本的IK分词器elasticsearch-analysis-ik-x.x.x.zip
在ES的plugins目录下创建ik目录 [giant@jd2 plugins]$ mkdir ik
将elasticsearch-analysis-ik-x.x.x.zip上传到plugins/ik目录下并解压 [giant@jd2 ik]$ unzip elasticsearch-analysis-ik-6.6.0.zip
删除elasticsearch-analysis-ik-x.x.x.zip安装包 [giant@jd2 ik]$ rm -rf elasticsearch-analysis-ik-6.6.0.zip [giant@jd2 ik]$ ll total 1428 -rw-r--r-- 1 giant giant 263965 Jan 15 17:07 commons-codec-1.9.jar -rw-r--r-- 1 giant giant 61829 Jan 15 17:07 commons-logging-1.2.jar drwxr-xr-x 2 giant giant 299 Jan 15 17:07 config -rw-r--r-- 1 giant giant 54693 Jan 15 17:07 elasticsearch-analysis-ik-6.6.0.jar -rw-r--r-- 1 giant giant 736658 Jan 15 17:07 httpclient-4.5.2.jar -rw-r--r-- 1 giant giant 326724 Jan 15 17:07 httpcore-4.4.4.jar -rw-r--r-- 1 giant giant 1805 Jan 15 17:07 plugin-descriptor.properties -rw-r--r-- 1 giant giant 125 Jan 15 17:07 plugin-security.policy
所有ES节点均进行以上操作，然后重启ES集群

IK分词器测试：

GET _analyze
{
  "analyzer": "ik_max_word",
  "text": "火箭明年总冠军"
}

{
  "tokens" : [
    {
      "token" : "火箭",
      "start_offset" : 0,
      "end_offset" : 2,
      "type" : "CN_WORD",
      "position" : 0
    },
    {
      "token" : "明年",
      "start_offset" : 2,
      "end_offset" : 4,
      "type" : "CN_WORD",
      "position" : 1
    },
    {
      "token" : "总冠军",
      "start_offset" : 4,
      "end_offset" : 7,
      "type" : "CN_WORD",
      "position" : 2
    },
    {
      "token" : "冠军",
      "start_offset" : 5,
      "end_offset" : 7,
      "type" : "CN_WORD",
      "position" : 3
    }
  ]
}

IK分词器有两种analyzer，ik_max_word和ik_smart

ik_max_word：会将文本做最细粒度的拆分
ik_smart：会做最粗粒度的拆分

3.4 IK分词器配置文件

IKAnalyzer.cfg.xml

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE properties SYSTEM "http://java.sun.com/dtd/properties.dtd">
<properties>
    <comment>IK Analyzer 扩展配置</comment>
    <!--用户可以在这里配置自己的扩展字典 -->
    <entry key="ext_dict"></entry>
     <!--用户可以在这里配置自己的扩展停止词字典-->
    <entry key="ext_stopwords"></entry>
    <!--用户可以在这里配置远程扩展字典 -->
    <!-- <entry key="remote_ext_dict">words_location</entry> -->
    <!--用户可以在这里配置远程扩展停止词字典-->
    <!-- <entry key="remote_ext_stopwords">words_location</entry> -->
</properties>

main.dic：IK分词器原生内置的中文词库，总共有27万多条，只要这里定义的单词，都会被分在一起
quantifier.dic：放了一些单位相关的词
suffix.dic：放了一些后缀单词
surname.dic：中国的姓氏
stopword.dic：英文停用词

3.5 自定义词库

自定义词库：每年都会涌现一些特殊的流行词，网红，蓝瘦香菇，喊麦，鬼畜，一般不会在ik的原生词典里，自己补充这些最新的词语，到ik的词库里面去，然后修改IKAnalyzer.cfg.xml配置文件
自定义停用词库：比如"了"，"的"，"啥"，"么"，我们可能并不想去建立索引，让人家搜索 <entry key="ext_dict">custom/mydict.dic</entry> <entry key="ext_stopwords">custom/mystopdict.dic</entry>
然后需要重启es，才能生效
测试

GET _analyze
{
  "analyzer": "ik_max_word",
  "text": "网红"
}

{
    "tokens": [
        {
            "token": "网",
            "start_offset": 0,
            "end_offset": 1,
            "type": "CN_CHAR",
            "position": 0
        },
        {
            "token": "红",
            "start_offset": 1,
            "end_offset": 2,
            "type": "CN_CHAR",
            "position": 1
        }
    ]
}

自定义词库

mkdir -p ${ELASTICSEARCH_HOME}/plugins/ik/config/custom
touch ${ELASTICSEARCH_HOME}/plugins/ik/config/custom/mydict.dic
# 然后把网红这个词写进去
# 然后修改IKAnalyzer.cfg.xml
<entry key="ext_dict">custom/mydict.dic</entry>

重启es，并测试

GET _analyze
{
  "analyzer": "ik_max_word",
  "text": "网红"
}

{
    "tokens": [
        {
            "token": "网红",
            "start_offset": 0,
            "end_offset": 2,
            "type": "CN_WORD",
            "position": 0
        }
    ]
}

本文参与腾讯云自媒体同步曝光计划，分享自作者个人站点/博客。

如有侵权请联系 cloudcommunity@tencent.com 删除

analyzer