Elasticsearch:构建自定义分析器指南

在本博客中,我们将介绍不同的内置字符过滤器、分词器和分词过滤器,以及如何创建适合我们需求的自定义分析器。更多关于分析器的知识,请详细阅读文章:

  • 开始使用 Elasticsearch (3)

  • Elasticsearch: analyzer

为什么我们需要定制分析器?

你可以通过以所需的方式组合字符过滤器、分词器和分词过滤器来创建自定义分析器来满足您的特定需求。 这使得文本处理具有高度的灵活性和定制性。

正如我们所见,Elasticsearch 中的分析器由三部分组成,我们将看到不同的内置组件:

安装

为了方便今天的测试,我们将安装无安全配置的 Elasticsearch 及 Kibana。我们可以参考文章 “Elasticsearch:如何在 Docker 上运行 Elasticsearch 8.x 进行本地开发”。

我们还需要安装 Python 所需要的包:

pip3 install elasticsearch
$ pip3 list | grep elasticsearch
elasticsearch                            8.12.0
rag-elasticsearch                        0.0.1        /Users/liuxg/python/rag-elasticsearch/my-app/packages/rag-elasticsearch

测试

我们创建一个连接到 Elasticsearch 的客户端:

from elasticsearch import Elasticsearches = Elasticsearch("http://localhost:9200")
print(es.info())

更多关于如何连接到 Elasticsearch 的代码,请参考 “Elasticsearch:关于在 Python 中使用 Elasticsearch 你需要知道的一切 - 8.x”。

Char map filters

HTML Strip Char Filter (html_strip)

从文本中删除 HTML 元素并解码 HTML 实体。

response = es.indices.analyze(body={"char_filter": ["html_strip"],"tokenizer": "standard","text": "<p>Hello <b>World</b>! This is <a href='<http://example.com>'>Elasticsearch</a>.</p>"}
)# Extract tokens
[token['token'] for token in response['tokens']]

Pattern Replace Char Filter (pattern_replace)

使用正则表达式来匹配字符或字符序列并替换它们。 在下面的示例中,我们从用户名中提取名称:

response = es.indices.analyze(body={"char_filter": [{"type": "pattern_replace","pattern": "[-_@.]",  # Removes hyphens, underscores, apostrophes"replacement": " "}],"tokenizer": "standard","text": "liu_xiao_guo"}
)# Extract tokens
[token['token'] for token in response['tokens']]

Mapping Char Filter (mapping)

允许自定义定义的字符或字符序列映射。 示例:你可以定义一个映射,将 “&” 替换为 “and”,或将 “€” 替换为 “euro”。

response = es.indices.analyze(body={"tokenizer": "standard","char_filter": [{"type": "mapping","mappings": ["@gmail.com=>",    # Replace @gmail.com with nothing"$=>dollar",       # Replace $ with dollar]}],"text": "xiaoguo.liu@gmail.com gives me $"}
)# Extract tokens
[token['token'] for token in response['tokens']]

Tokenizers

Standard Tokenizer (standard)

Standard 分词器将文本按照单词边界划分为术语,如 Unicode 文本分段算法所定义。 它删除了大多数标点符号。 它是大多数语言的最佳选择。

response = es.indices.analyze(body={"tokenizer": "standard","text": "The 2 QUICK Brown-Foxes, jumps_over the lazy-dog's bone."}
)# Extract tokens
[token['token'] for token in response['tokens']]

Letter Tokenizer (letter)

每当遇到非字母的字符时,letter 分词器就会将文本分成术语。

response = es.indices.analyze(body={"tokenizer": "letter","text": "The 2 QUICK Brown-Foxes, jumps_over the lazy-dog's bone."}
)# Extract tokens
[token['token'] for token in response['tokens']]

Lowercase Tokenizer (lowercase)

小写分词器类似于字母分词器,但它也将所有术语小写。

response = es.indices.analyze(body={"tokenizer": "lowercase","text": "The 2 QUICK Brown-Foxes, jumps_over the lazy-dog's bone."}
)# Extract tokens
[token['token'] for token in response['tokens']]

Whitespace Tokenizer (whitespace)

每当遇到任何空白字符时,whitespace 分词器都会将文本划分为术语。

response = es.indices.analyze(body={"tokenizer": "whitespace","text": "The 2 QUICK Brown-Foxes, jumps_over the lazy-dog's bone."}
)# Extract tokens
[token['token'] for token in response['tokens']]

Classic Tokenizer (classic)

classic 分词器是一种基于语法的英语分词器。

response = es.indices.analyze(body={"tokenizer": "classic","text": "The 2 QUICK Brown-Foxes, jumps_over the lazy-dog's bone."}
)# Extract tokens
[token['token'] for token in response['tokens']]

UAX URL Email Tokenizer (uax_url_email)

uax_url_email 标记生成器类似于标准标记生成器,只不过它将 URL 和电子邮件地址识别为单个标记。

response = es.indices.analyze(body={"tokenizer": "classic","text": "visit https://elasticstack.blog.csdn.net to get the best materials to learn Elastic Stack"}
)# Extract tokens
[token['token'] for token in response['tokens']]

N-Gram Tokenizer (ngram)

当 ngram 分词器遇到任何指定字符(例如空格或标点符号)列表时,它可以将文本分解为单词,然后返回每个单词的 n-grams:连续字母的滑动窗口,例如 Quick → [qu, ui, ic, ck]。Elasticsearch 中的 N-Gram 分词器在术语部分匹配很重要的场景中特别有用。 最适合自动完成和键入时搜索功能以及处理拼写错误或匹配单词中的子字符串。

response = es.indices.analyze(body={"tokenizer": {"type": "ngram","min_gram": 3,"max_gram": 4},"text": "Hello Xiaoguo"}
)# Extract tokens
[token['token'] for token in response['tokens']]

Edge N-Gram Tokenizer (edge_ngram)

Elasticsearch 中的 edge_ngram 分词器用于从单词的开头或 “边缘” 开始将单词分解为更小的块或 “n-gram”。 它生成指定长度范围的标记,提供单词从开头到给定大小的一部分。

response = es.indices.analyze(body={"tokenizer": {"type": "edge_ngram","min_gram": 4,"max_gram": 5,"token_chars": ["letter", "digit"]},"text": "The 2 QUICK Brown-Foxes, jumps_over the lazy-dog's bone."}
)# Extract tokens
[token['token'] for token in response['tokens']]

Keyword Tokenizer (keyword)

关键字分词器接受给定的任何文本,并将完全相同的文本输出为单个术语。

response = es.indices.analyze(body={"tokenizer": "keyword","text": "The 2 QUICK Brown-Foxes, jumps_over the lazy-dog's bone."}
)# Extract tokens
[token['token'] for token in response['tokens']]

Pattern Tokenizer (pattern)

Pattern 分词器使用正则表达式,在文本与单词分隔符匹配时将其拆分为术语,或者将匹配的文本捕获为术语。

response = es.indices.analyze(body={"tokenizer": {"type": "pattern","pattern": "_+"},"text": "hello_world_from_elasticsearch"}
)# Extract tokens
[token['token'] for token in response['tokens']]

Path Tokenizer (path_hierarchy)

它将路径在每个路径分隔符处分解为分词。

response = es.indices.analyze(body={"tokenizer": "path_hierarchy","text": "/usr/local/bin/python"}
)# Extract tokens
[token['token'] for token in response['tokens']]

Token filters

确保你始终传递列表中的过滤器,即使它只有一个,并且你应用的过滤器的顺序非常重要。

Apostrophe

删除撇号后面的所有字符,包括撇号本身。

response = es.indices.analyze(body={"filter": ["apostrophe"],"tokenizer": "standard","text": "The 2 QUICK Brown-Foxes, jumps_over the lazy-dog's bone."}
)# Extract tokens
[token['token'] for token in response['tokens']]

Lowercase Filter

将所有分词转换为小写。

response = es.indices.analyze(body={"filter": ["lowercase"],"text": "The 2 QUICK Brown-Foxes, jumps_over the lazy-dog's bone."}
)# Extract tokens
[token['token'] for token in response['tokens']]

Uppercase Filter

将所有分词转换为大写。

response = es.indices.analyze(body={"filter": ["uppercase"],"text": "The 2 QUICK Brown-Foxes, jumps_over the lazy-dog's bone."}
)# Extract tokens
[token['token'] for token in response['tokens']]

Trim Filter

删除流中每个分词的前导和尾随空格。

# Analyze the text using the custom analyzer
response = es.indices.analyze(body={"tokenizer": "keyword","filter":["lowercase","trim"],"text": " The 2 QUICK Brown-Foxes, jumps_over the lazy-dog's bone. "}
)# Extract tokens
[token['token'] for token in response['tokens']]

ASCII Folding Filter (asciifolding)

asciifolding 过滤器会删除标记中的变音标记。比如,Türkiye 将成为 Turkiye。

# Analyze the text using the custom analyzer
response = es.indices.analyze(body={"filter": ["asciifolding"],"text": "Türkiye"}
)# Extract tokens
[token['token'] for token in response['tokens']]

Synonym Filter

synonym 分词过滤器允许在分析过程中轻松处理同义词。

# Analyze the text using the custom analyzer
response = es.indices.analyze(body={"tokenizer": "standard","filter":["lowercase",{"type": "synonym","synonyms": ["jumps_over => leap"]}],"text": "The 2 QUICK Brown-Foxes, jumps_over the lazy-dog's bone."}
)# Extract tokens
[token['token'] for token in response['tokens']]

Synonym Graph Filter

最适合多词同义词。

response = es.indices.analyze(body={"tokenizer": "standard","filter":["lowercase",{"type": "synonym_graph","synonyms": ["NYC, New York City", "LA, Los Angeles"]}],"text": "Flight from LA to NYC has been delayed by an hour"}
)# Extract tokens
[token['token'] for token in response['tokens']]

请记住,输出并不直观地表示内部图形结构,但 Elasticsearch 在搜索查询期间使用此结构。

与通常同义词不匹配的匹配短语查询将与同义词图完美配合。

Stemmer Filter

词干过滤器,支持多种语言的词干提取。

response = es.indices.analyze(body={"tokenizer": "standard","filter": [{"type": "stemmer","language": "English",},],"text": "candies, ladies, plays, playing, ran, running, dresses"}
)# Extract tokens
[token['token'] for token in response['tokens']]

KStem Filter

kstem 过滤器将算法词干提取与内置字典相结合。 与其他英语词干分析器(例如 porter_stem 过滤器)相比,kstem 过滤器的词干提取力度较小。

response = es.indices.analyze(body={"tokenizer": "standard","filter": ['kstem',],"text": "candies, ladies, plays, playing, ran, running"}
)# Extract tokens
[token['token'] for token in response['tokens']]

Porter Stem Filter

与其他英语词干过滤器(例如 kstem 过滤器)相比,倾向于更积极地进行词干提取。

response = es.indices.analyze(body={"tokenizer": "whitespace","filter": [{"type": "pattern_replace","pattern": "[-|.|,]"},{"type": "porter_stem","language": "English",},],"text": "candies, ladies, plays, playing, ran, running, dresses"}
)# Extract tokens
[token['token'] for token in response['tokens']]

Snowball Filter

使用 Snowball 生成的词干分析器对单词进行词干分析的过滤器。 适用于法语、德语、俄语、西班牙语等不同语言。

response = es.indices.analyze(body={"tokenizer": "whitespace","filter": [{"type": "snowball","language": "English",},],"text": "candies, ladies, plays, playing, ran, running, dresses"}
)# Extract tokens
[token['token'] for token in response['tokens']]

Stemmer Override

通过应用自定义映射来覆盖词干算法,然后保护这些术语不被词干分析器修改。 必须放置在任何阻塞过滤器之前。

response = es.indices.analyze(body={"tokenizer": "standard","filter": [{"type": "stemmer_override","language": "English","rules": ["running, runs => run","stemmer => stemmer"]},],"text": "candies, ladies, plays, playing, ran, running, dresses"}
)# Extract tokens
[token['token'] for token in response['tokens']]

更多使用方法,请参考 Stemmer override token filter | Elasticsearch Guide [8.12] | Elastic

Keyword Marker Filter

将某些术语标记为关键字,防止它们被其他过滤器(如词干分析器)修改。

response = es.indices.analyze(body={"tokenizer": "whitespace","filter": [{"type": "keyword_marker","keywords": ["running"]  # Mark 'running' as a keyword},{"type": "pattern_replace","pattern": "[-|.|,]"},{"type": "porter_stem","language": "English",},],"text": "candies, ladies, plays, playing, runs, running"}
)# Extract tokens
[token['token'] for token in response['tokens']]

Stop Filter

从分词流中删除停用词(经常被忽略的常用词)。 示例 — if、of、is、am、are、the。可以使用默认或自定义的停用词列表。

# Analyze the text using the custom analyzer
response = es.indices.analyze(body={"tokenizer": "standard","filter":{"type":"stop","stopwords": ["is","am","are","of","if","a","the"],"ignore_case": True},"text": "i am sachin. I Am software engineer."}
)# Extract tokens
[token['token'] for token in response['tokens']]

Unique Filter

从流中删除重复的分词。

response = es.indices.analyze(body={"tokenizer": "whitespace","filter":["lowercase", "unique",],"text": "Happy happy joy joy"}
)# Extract tokens
[token['token'] for token in response['tokens']]

Length Filter

删除比指定字符长度更短或更长的分词。

response = es.indices.analyze(body={"tokenizer": "standard","filter":["lowercase",{"type": "length","min": 1,"max": 4}],"text": "The 2 QUICK Brown-Foxes, jumps_over the lazy-dog's bone."}
)# Extract tokens
[token['token'] for token in response['tokens']]

NGram Token Filter

从分词形成指定长度的 ngram。 最适合在键入时自动完成或搜索。 或者用于用户可能会犯错或拼写错误的搜索。

response = es.indices.analyze(body={"tokenizer": "whitespace","filter":[{"type": "ngram","min_gram": 3,"max_gram": 4}],"text": "Skinny blue jeans by levis"}
)# Extract tokens
[token['token'] for token in response['tokens']]

Edge NGram Token Filter

从分词的开头形成指定长度的 ngram。 最适合在键入时自动完成或搜索。 它对于搜索建议中常见的部分单词匹配非常有效。

response = es.indices.analyze(body={"tokenizer": "whitespace","filter":[{"type": "edge_ngram","min_gram": 3,"max_gram": 4}],"text": "Skinny blue jeans by levis"}
)# Extract tokens
[token['token'] for token in response['tokens']]

Shingle Filter

通过连接相邻的标记,将 shingles 或单词 ngram 添加到分词流中。 默认情况下,shingle 分词过滤器输出两个字的 shingles。 最适用于提高搜索短语查询性能。

response = es.indices.analyze(body={"tokenizer": "whitespace","filter":[{"type": "shingle","min_shingle_size": 2,"max_shingle_size": 3           }],"text": "Welcome to use Elastic Stack"}
)[token['token'] for token in response['tokens']]

Creating a custom analyzer

以下是文本,下面是所需的输出:

text = "The 2 QUICK Brown-Foxes, jumps_over the lazy-dog's bone."# Desired output
['2', 'quick', 'brown', 'fox', 'jump', 'over', 'lazy', 'dog', 'bone']

分析器应完成的事情列表:

  • 删除所有符号 - 连字符和下划线。
  • 删除停用词。
  • 将所有文本小写。
  • 删除撇号。
  • 词干。
response = es.indices.analyze(body={"char_filter": [{"type": "mapping","mappings": ["- => ' '", # replacing hyphens with blank space"_ => ' '", # replacing underscore with blank space]}],"tokenizer": "standard","filter": ["apostrophe", "lowercase", "stop", "porter_stem"],"text": "The 2 QUICK Brown-Foxes, jumps_over the lazy-dog's bone."}
)# Extract and print tokens
tokens = [token['token'] for token in response['tokens']]
tokens

现在需要注意的一件事是顺序,无论你在内部处理时给 Elasticsearch 什么顺序,总是使用相同的顺序 char_filter > tokenizer > token_filter 但 char_filter 或 token filter 块内的顺序会有所不同。

将自定义分析器添加到索引

为了避免复杂化,最好创建一个新的索引并根据你的要求设置分析器。 以下是设置分析器的方法。

settings = {"settings": {"analysis": {"analyzer": {"my_custom_analyzer": {"type": "custom","char_filter": {"type": "mapping","mappings": ["- => ' '","_ => ' '",]},"tokenizer": "standard","filter": ["lowercase", "apostrophe", "stop", "porter_stem"],}}},"index": {"number_of_shards": 1,"number_of_replicas": 0,"routing.allocation.include._tier_preference": "data_hot"},},"mappings": {"properties": {"title": {"type":"text", "analyzer":"my_custom_analyzer"},"brand": {"type": "text", "analyzer":"my_custom_analyzer", "fields": {"raw": {"type": "keyword"}}},"updated_time": {"type": "date"}}`}
}response = es.indices.create(index="trial_index", body=index_settings)

你可以在地址找到所有的代码:https://github.com/liu-xiao-guo/analyzers-python

本文来自互联网用户投稿,该文观点仅代表作者本人,不代表本站立场。本站仅提供信息存储空间服务,不拥有所有权,不承担相关法律责任。如若转载,请注明出处:http://www.hqwc.cn/news/442781.html

如若内容造成侵权/违法违规/事实不符,请联系编程知识网进行投诉反馈email:809451989@qq.com,一经查实,立即删除!

相关文章

【爬虫专区】批量下载PDF (无反爬)

天命:只要没反爬,一切都简单 这次爬取的是绿盟的威胁情报的PDF 先看一下结构,很明显就是一个for循环渲染 burp抓包会发现第二次接口请求 接口请求一次就能获取到了所有的数据 然后一个循环批量下载数据即可,其实没啥难度的 import requests,osres = requests.get("…

力扣 55.跳跃游戏

思路&#xff1a; 从后往前遍历&#xff0c;遇到元素为0时&#xff0c;记录对应的下标位置&#xff0c;再向前遍历元素&#xff0c;看最大的跳跃步数能否跳过0的位置&#xff0c;不能则继续往前遍历 代码&#xff1a; class Solution { public:bool canJump(vector<int>…

【C语言】探索数据结构:单链表和双链表

目录 &#x1f4a1;链表的概念和结构 &#x1f4a1;链表的分类 &#x1f4a1;无头单向非循环链表&#xff08;单链表&#xff09;的实现 定义节点结构 单链表的尾部插入 单链表的头部插入 单链表的尾部删除 单链表的头部删除 在指定位置插入前数据 在指定位置之后插入数…

nodejs+vue+ElementUi大学生校园生活互助系统nyvlx

本文讲述了大学生爱心互助代购网站。结合电子管理系统的特点&#xff0c;分析了大学生爱心互助代购网站的背景&#xff0c;给出了大学生爱心互助代购网站实现的设计方案。 本论文主要完成不同用户的权限划分&#xff0c;不同用户具有不同权限的操作功能&#xff0c;在学生模块&…

Centos 7.9 在线安装 VirtualBox 7.0

1 访问 Linux_Downloads – Oracle VM VirtualBox 2 点击 ​the Oracle Linux repo file 复制 内容到 /etc/yum.repos.d/. 3 在 /etc/yum.repos.d/ 目录下新建 virtualbox.repo&#xff0c;复制内容到 virtualbox.repo 并 :wq 保存。 [rootlocalhost centos]# cd /etc/yum.rep…

聚观早报 | 360 AI搜索App上线;岚图汽车与京东达成合作

聚观早报每日整理最值得关注的行业重点事件&#xff0c;帮助大家及时了解最新行业动态&#xff0c;每日读报&#xff0c;就读聚观365资讯简报。 整理丨Cutie 1月30日消息 360 AI搜索App上线 岚图汽车与京东达成合作 三星电子在硅谷新设实验室 小米平板7系列参数曝光 Spa…

sql注入,布尔盲注和时间盲注,无回显

布尔盲注 通过order by分组可以看到&#xff0c;如果正确会i显示you are in&#xff0c;错误则无任何提示&#xff0c;由此可以判断出&#xff0c;目前只显示对错&#xff0c;此外前端不会显示任何数据 也就是说&#xff0c;目前结果只有两种&#xff0c;在这种只有两种变量的…

大数据信用报告应该去哪里查询比较好呢?

对于个人而言&#xff0c;大数据信用报告也变得越来越重要。那么&#xff0c;大数据信用报告应该去哪里查询呢?本文将为您详细介绍征信和大数据的区别&#xff0c;并推荐一个可靠的大数据平台。 首先&#xff0c;我们需要了解征信和大数据的区别 征信报告 依法采集、整理、保存…

力扣1446连续字符

题解&#xff1a; 1、初始化两个变量&#xff1a;ans 和 num&#xff0c;分别设置为1。ans 用于存储最长连续重复字符的长度&#xff0c;而 num 用于计算当前连续重复字符的长度。 2、遍历字符串 s&#xff0c;从索引1开始&#xff08;因为索引0没有前一个字符可以比较&#x…

科技感十足的Pencil平替,功能全面手感丝滑,西圣Pencil 2上手

搭配Apple Pencil的iPad的确实可以大大提升工作效率&#xff0c;但是原厂的Apple Pencil价格实在偏高&#xff0c;而且容易遗失&#xff0c;所以很多人都会选择一些Apple Pencil的平替。最近我在用一款西圣Pencil 2&#xff0c;这款电容笔设计很有特点&#xff0c;看起来科技感…

Prometheus结合Consul采集多个MySQL实例的监控指标

本文主要介绍如何利用Prometheus官网提供的mysqld_exporter进行多MySQL实例的监控指标采集 建议安装最新版的mysqld_exporter&#xff0c; 因为从’2022-09-01’ 之后才支持多实例功能的。具体的官网说明详见 Support for scraping multiple mysqld hosts (#651) ok&#xff0…

C51 单片机学习(一):基础外设

参考 51单片机入门教程 1. 单片机简介 1.1 定义 单片机&#xff08;Micro Controller Unit&#xff0c;简称 MCU&#xff09; 内部集成了 CPU、RAM、ROM、定时器、中断系统、通讯接口等一系列电脑的常用硬件功能单片机的任务是信息采集&#xff08;依靠传感器&#xff09;、处…