Convert Figma logo to code with AI

chatopera logoSynonyms

:herb: 中文近义词：聊天机器人，智能问答工具包

5,017
904
5,017
30
View on GitHub

Top Related Projects

fxsjy's avatar

jieba

33,199

结巴中文分词

hankcs's avatar

HanLP

33,691

中文分词 词性标注 命名实体识别 依存句法分析 成分句法分析 语义依存分析 语义角色标注 指代消解 风格转换 语义相似度 新词发现 关键词短语提取 自动摘要 文本分类聚类 拼音简繁转换 自然语言处理

isnowfy's avatar

snownlp

6,411

Python library for processing Chinese text

baidu's avatar

lac

3,863

百度NLP：分词，词性标注，命名实体识别，词重要性

bojone's avatar

bert4keras

5,360

keras implement of transformers for humans

Quick Overview

Synonyms is a Chinese Natural Language Processing (NLP) library for word similarity and sentence similarity calculations. It provides tools for semantic analysis, word embedding, and related NLP tasks specifically tailored for the Chinese language.

Pros

  • Specialized for Chinese language processing
  • Offers both word and sentence similarity calculations
  • Includes pre-trained word vectors for immediate use
  • Supports custom word vectors and dictionaries

Cons

  • Limited documentation in English
  • May require additional resources for optimal performance
  • Focused primarily on similarity calculations, lacking broader NLP features
  • Relatively small community compared to more general-purpose NLP libraries

Code Examples

  1. Calculate word similarity:
import synonyms
r = synonyms.compare('北京', '上海')
print(r)  # Output: 0.63395087
  1. Find synonyms for a given word:
import synonyms
words, scores = synonyms.nearby('北京')
print(words[:3])  # Output: ['首都', '城市', '中国']
print(scores[:3])  # Output: [0.896611, 0.754657, 0.74305]
  1. Calculate sentence similarity:
import synonyms
sen1 = '我喜欢吃苹果'
sen2 = '我喜欢吃香蕉'
r = synonyms.compare(sen1, sen2, seg=True)
print(r)  # Output: 0.9285714285714286

Getting Started

To get started with Synonyms:

  1. Install the library:
pip install synonyms
  1. Import and use in your Python script:
import synonyms

# Calculate word similarity
similarity = synonyms.compare('北京', '上海')
print(f"Similarity between '北京' and '上海': {similarity}")

# Find synonyms
words, scores = synonyms.nearby('学习')
print(f"Synonyms for '学习': {words[:5]}")
print(f"Scores: {scores[:5]}")

Note: Make sure you have sufficient disk space (about 1.5GB) for the pre-trained word vectors.

Competitor Comparisons

fxsjy logo

jieba

33,199

结巴中文分词

Pros of jieba

  • More mature and widely adopted Chinese text segmentation library
  • Supports multiple segmentation modes (accurate, full, search engine)
  • Extensive documentation and community support

Cons of jieba

  • Primarily focused on word segmentation, not synonyms
  • Less emphasis on semantic analysis and word relationships

Code Comparison

jieba:

import jieba

seg_list = jieba.cut("我来到北京清华大学", cut_all=False)
print("Default Mode: " + "/ ".join(seg_list))

Synonyms:

import synonyms

words = synonyms.seg("我来到北京清华大学")
print(words)

Key Differences

  • jieba is primarily a word segmentation tool, while Synonyms focuses on both segmentation and synonym generation
  • Synonyms provides word vectors and similarity calculations, which are not core features of jieba
  • jieba offers more granular control over segmentation modes, while Synonyms emphasizes semantic understanding

Use Cases

  • jieba: Best for applications requiring precise Chinese word segmentation
  • Synonyms: Ideal for projects needing both segmentation and semantic analysis, such as text similarity comparisons or synonym suggestions

Community and Maintenance

  • jieba: Larger user base, more frequent updates, and extensive third-party integrations
  • Synonyms: Smaller but growing community, with a focus on semantic analysis and NLP applications
hankcs logo

HanLP

33,691

中文分词 词性标注 命名实体识别 依存句法分析 成分句法分析 语义依存分析 语义角色标注 指代消解 风格转换 语义相似度 新词发现 关键词短语提取 自动摘要 文本分类聚类 拼音简繁转换 自然语言处理

Pros of HanLP

  • More comprehensive NLP toolkit with broader functionality
  • Better support for traditional Chinese characters
  • More active development and frequent updates

Cons of HanLP

  • Steeper learning curve due to more complex API
  • Larger library size, potentially impacting performance
  • Requires more setup and configuration

Code Comparison

HanLP:

from pyhanlp import *

text = "我爱北京天安门"
print(HanLP.segment(text))

Synonyms:

import synonyms

words = synonyms.seg("我爱北京天安门")
print(words)

Both libraries offer word segmentation functionality, but HanLP provides more detailed output with part-of-speech tagging. Synonyms focuses primarily on word similarity and segmentation, while HanLP offers a wider range of NLP tasks.

HanLP is better suited for projects requiring advanced NLP capabilities in Chinese, including named entity recognition, dependency parsing, and more. Synonyms is more appropriate for simpler tasks focused on word relationships and basic segmentation.

Consider your project's specific needs, performance requirements, and the level of NLP functionality required when choosing between these libraries.

isnowfy logo

snownlp

6,411

Python library for processing Chinese text

Pros of SnowNLP

  • Broader functionality including sentiment analysis, text classification, and word segmentation
  • Includes tools for pinyin conversion and simplified/traditional Chinese conversion
  • More comprehensive documentation and examples

Cons of SnowNLP

  • Less focused on synonyms and semantic similarity
  • May require more setup and configuration for specific tasks
  • Not as actively maintained (last update was in 2020)

Code Comparison

SnowNLP example:

from snownlp import SnowNLP

s = SnowNLP(u'这个东西真心很赞')
print(s.sentiments)  # Sentiment analysis
print(s.pinyin)  # Pinyin conversion

Synonyms example:

import synonyms

print(synonyms.nearby('人脸'))
print(synonyms.compare('北京', '上海', seg=True))

Summary

SnowNLP offers a wider range of NLP functionalities for Chinese text processing, including sentiment analysis and text classification. However, it's less focused on synonyms and semantic similarity compared to Synonyms. SnowNLP provides more comprehensive documentation but hasn't been updated as recently as Synonyms. The choice between the two depends on the specific NLP tasks required for your project.

baidu logo

lac

3,863

百度NLP：分词，词性标注，命名实体识别，词重要性

Pros of LAC

  • Offers comprehensive Chinese language processing capabilities, including word segmentation, part-of-speech tagging, and named entity recognition
  • Provides pre-trained models for various domains, enhancing accuracy and performance
  • Supports both Python and C++ interfaces, offering flexibility for different development environments

Cons of LAC

  • Primarily focused on Chinese language processing, limiting its applicability for other languages
  • May require more computational resources due to its comprehensive feature set
  • Has a steeper learning curve compared to simpler synonym-focused libraries

Code Comparison

LAC:

from LAC import LAC

lac = LAC(mode='lac')
text = "LAC是个优秀的中文处理工具"
result = lac.run(text)
print(result)

Synonyms:

import synonyms

word = "优秀"
synonyms = synonyms.nearby(word)
print(synonyms)

Key Differences

LAC is a more comprehensive Chinese language processing tool, offering a wide range of features beyond synonym detection. It's well-suited for complex NLP tasks in Chinese. Synonyms, on the other hand, is more focused on providing synonym functionality and is simpler to use for basic word similarity tasks. LAC may be preferred for large-scale Chinese NLP projects, while Synonyms could be more appropriate for quick synonym lookups or simpler language processing needs.

bojone logo

bert4keras

5,360

keras implement of transformers for humans

Pros of bert4keras

  • More comprehensive and flexible BERT implementation
  • Supports multiple BERT variants and architectures
  • Better suited for advanced NLP tasks beyond synonyms

Cons of bert4keras

  • Steeper learning curve and more complex setup
  • Requires more computational resources
  • May be overkill for simple synonym-related tasks

Code Comparison

Synonyms:

import synonyms
synonyms.nearby("你好")

bert4keras:

from bert4keras.models import build_transformer_model
from bert4keras.tokenizers import Tokenizer

model = build_transformer_model(config_path, checkpoint_path)
tokenizer = Tokenizer(dict_path)

Summary

Synonyms is a lightweight library focused specifically on Chinese synonym detection and word similarity. It's easy to use and suitable for simple NLP tasks. bert4keras, on the other hand, is a more powerful and versatile BERT implementation that can handle a wide range of NLP tasks but requires more setup and resources. Choose Synonyms for quick synonym-related projects, and bert4keras for more complex NLP applications requiring BERT's capabilities.

Convert Figma logo designs to code with AI

Visual Copilot

Introducing Visual Copilot: A new AI model to turn Figma designs to high quality code using your components.

Try Visual Copilot

README

PyPI PyPI download month License

Synonyms

Chinese Synonyms for Natural Language Processing and Understanding.

更好的中文近义词：聊天机器人、智能问答工具包。

synonyms可以用于自然语言理解的很多任务：文本对齐，推荐算法，相似度计算，语义偏移，关键字提取，概念提取，自动摘要，搜索引擎等。

为提供稳定、可靠、长期优化的服务，Synonyms 改为使用 春松许可证, v1.0 并针对机器学习模型的下载进行收费，详见证书商店。之前的贡献者（突出贡献的代码贡献者），可与我们联系，讨论收费问题。-- Chatopera Inc. @ Oct. 2023

Table of Content:

Welcome

Follow steps below to install and activate packages.

1/3 Install Sourcecodes Package

pip install -U synonyms

当前稳定版本 v3.x。

2/3 Config license id

Synonyms's machine learning model package(s) requires a License from Chatopera License Store, first purchase a License and get the license id from Licenses page on Chatopera License Store(license id：在证书商店，证书详情页，点击【复制证书标识】).

image

Secondly, set environment variable in your terminal or shell scripts as below.

  • For Shell Users

e.g. Shell, CMD Scripts on Linux, Windows, macOS.

# Linux / macOS
export SYNONYMS_DL_LICENSE=YOUR_LICENSE
## e.g. if your license id is `FOOBAR`, run `export SYNONYMS_DL_LICENSE=FOOBAR`

# Windows
## 1/2 Command Prompt
set SYNONYMS_DL_LICENSE=YOUR_LICENSE
## 2/2 PowerShell
$env:SYNONYMS_DL_LICENSE='YOUR_LICENSE'
  • For Python Code Users

Jupyter Notebook, etc.

import os
os.environ["SYNONYMS_DL_LICENSE"] = "YOUR_LICENSE"
_licenseid = os.environ.get("SYNONYMS_DL_LICENSE", None)
print("SYNONYMS_DL_LICENSE=", _licenseid)

提示：安装后初次使用会下载词向量文件，下载速度取决于网络情况。

3/3 Download Model Package

Last, download the model package by command or script -

python -c "import synonyms; synonyms.display('能量')" # download word vectors file

Usage

支持使用环境变量配置分词词表和 word2vec 词向量文件。

环境变量描述
SYNONYMS_WORD2VEC_BIN_MODEL_ZH_CN使用 word2vec 训练的词向量文件，二进制格式。
SYNONYMS_WORDSEG_DICT中文分词主字典，格式和使用参考
SYNONYMS_DEBUG["TRUE"|"FALSE"], 是否输出调试日志，设置为 “TRUE” 输出，默认为 “FALSE”

synonyms#nearby(word [, size = 10])

import synonyms
print("人脸: ", synonyms.nearby("人脸"))
print("识别: ", synonyms.nearby("识别"))
print("NOT_EXIST: ", synonyms.nearby("NOT_EXIST"))

synonyms.nearby(WORD [,SIZE])返回一个元组，元组中包含两项：([nearby_words], [nearby_words_score])，nearby_words是 WORD 的近义词们，也以 list 的方式存储，并且按照距离的长度由近及远排列，nearby_words_score是nearby_words中**对应位置**的词的距离的分数，分数在(0-1)区间内，越接近于 1，代表越相近；SIZE 是返回词汇数量，默认 10。比如:

synonyms.nearby(人脸, 10) = (
    ["图片", "图像", "通过观察", "数字图像", "几何图形", "脸部", "图象", "放大镜", "面孔", "Mii"],
    [0.597284, 0.580373, 0.568486, 0.535674, 0.531835, 0.530
095, 0.525344, 0.524009, 0.523101, 0.516046])

在 OOV 的情况下，返回 ([], [])，目前的字典大小: 435,729。

synonyms#compare(sen1, sen2 [, seg=True])

两个句子的相似度比较

    sen1 = "发生历史性变革"
    sen2 = "发生历史性变革"
    r = synonyms.compare(sen1, sen2, seg=True)

其中，参数 seg 表示 synonyms.compare 是否对 sen1 和 sen2 进行分词，默认为 True。返回值：[0-1]，并且越接近于 1 代表两个句子越相似。

旗帜引领方向 vs 道路决定命运: 0.429
旗帜引领方向 vs 旗帜指引道路: 0.93
发生历史性变革 vs 发生历史性变革: 1.0

synonyms#display(word [, size = 10])

以友好的方式打印近义词，方便调试，display(WORD [, SIZE])调用了 synonyms#nearby 方法。

>>> synonyms.display("飞机")
'飞机'近义词：
  1. 飞机:1.0
  2. 直升机:0.8423391
  3. 客机:0.8393003
  4. 滑翔机:0.7872388
  5. 军用飞机:0.7832081
  6. 水上飞机:0.77857226
  7. 运输机:0.7724742
  8. 航机:0.7664748
  9. 航空器:0.76592904
  10. 民航机:0.74209654

SIZE 是打印词汇表的数量，默认 10。

synonyms#describe()

打印当前包的描述信息：

>>> synonyms.describe()
Vocab size in vector model: 435729
model_path: /Users/hain/chatopera/Synonyms/synonyms/data/words.vector.gz
version: 3.18.0
{'vocab_size': 435729, 'version': '3.18.0', 'model_path': '/chatopera/Synonyms/synonyms/data/words.vector.gz'}

synonyms#v(word)

获得一个词语的向量，该向量为 numpy 的 array，当该词语是未登录词时，抛出 KeyError 异常。

>>> synonyms.v("飞机")
array([-2.412167  ,  2.2628384 , -7.0214124 ,  3.9381874 ,  0.8219283 ,
       -3.2809453 ,  3.8747153 , -5.217062  , -2.2786229 , -1.2572327 ],
      dtype=float32)

synonyms#sv(sentence, ignore=False)

获得一个分词后句子的向量，向量以 BoW 方式组成

    sentence: 句子是分词后通过空格联合起来
    ignore: 是否忽略OOV，False时，随机生成一个向量

synonyms#seg(sentence)

中文分词

synonyms.seg("中文近义词工具包")

分词结果，由两个 list 组成的元组，分别是单词和对应的词性。

(['中文', '近义词', '工具包'], ['nz', 'n', 'n'])

该分词不去停用词和标点。

synonyms#keywords(sentence [, topK=5, withWeight=False])

提取关键词，默认按照重要程度提取关键词。

keywords = synonyms.keywords("9月15日以来，台积电、高通、三星等华为的重要合作伙伴，只要没有美国的相关许可证，都无法供应芯片给华为，而中芯国际等国产芯片企业，也因采用美国技术，而无法供货给华为。目前华为部分型号的手机产品出现货少的现象，若该形势持续下去，华为手机业务将遭受重创。")

Contribution

Get more logs for debugging, set environment variable.

SYNONYMS_DEBUG=TRUE

PCA

以“人脸”为例主要成分分析：

Quick Get Start

$ pip install -r Requirements.txt
$ python demo.py

Change logs

更新情况说明。

Voice of Users

用户怎么说：

Data

data is built based on wikidata-corpus.

Valuation

同义词词林

《同义词词林》是梅家驹等人于 1983 年编纂而成，现在使用广泛的是哈工大社会计算与信息检索研究中心维护的《同义词词林扩展版》，它精细的将中文词汇划分成大类和小类，梳理了词汇间的关系，同义词词林扩展版包含词语 7 万余条，其中 3 万余条被以开放数据形式共享。

知网, HowNet

HowNet，也被称为知网，它并不只是一个语义字典，而是一个知识系统，词汇之间的关系是其一个基本使用场景。知网包含词语 8 余条。

国际上对词语相似度算法的评价标准普遍采用 Miller&Charles 发布的英语词对集的人工判定值。该词对集由十对高度相关、十对中度相关、十对低度相关共 30 个英语词对组成,然后让 38 个受试者对这 30 对进行语义相关度判断，最后取他们的平均值作为人工判定标准。然后不同近义词工具也对这些词汇进行相似度评分，与人工判定标准做比较，比如使用皮尔森相关系数。在中文领域，使用这个词表的翻译版进行中文近义词比较也是常用的办法。

对比

Synonyms 的词表容量是 435,729，下面选择一些在同义词词林、知网和 Synonyms 都存在的几个词，给出其近似度的对比：

注：同义词林及知网数据、分数来源。Synonyms 也在不断优化中，新的分数可能和上图不一致。

更多比对结果。

Used by

Github 关联用户列表

Benchmark

Test with py3, MacBook Pro.

python benchmark.py

++++++++++ OS Name and version ++++++++++

Platform: Darwin

Kernel: 16.7.0

Architecture: ('64bit', '')

++++++++++ CPU Cores ++++++++++

Cores: 4

CPU Load: 60

++++++++++ System Memory ++++++++++

meminfo 8GB

synonyms#nearby: 100000 loops, best of 3 epochs: 0.209 usec per loop

Live Sharing

52nlp.cn

机器之心

线上分享实录: Synonyms 中文近义词工具包 @ 2018-02-07

Statement

Synonyms发布证书 MIT。数据和程序可用于研究和商业产品，必须注明引用和地址，比如发布的任何媒体、期刊、杂志或博客等内容。

@online{Synonyms:hain2017,
  author = {Hai Liang Wang, Hu Ying Xi},
  title = {中文近义词工具包Synonyms},
  year = 2017,
  url = {https://github.com/chatopera/Synonyms},
  urldate = {2017-09-27}
}

References

wikidata-corpus

word2vec 原理推导与代码分析

Frequently Asked Questions (FAQ)

  1. 是否支持添加单词到词表中？

不支持，欲了解更多请看 #5

  1. 词向量的训练是用哪个工具？

Google 发布的word2vec，该库由 C 语言编写，内存使用效率高，训练速度快。gensim 可以加载 word2vec 输出的模型文件。

  1. 相似度计算的方法是什么？

详见 #64

  1. #118 词向量文件一直下载不下来？

Authors

Hai Liang Wang

Hu Ying Xi

自然语言处理推荐入门&工具书

本书由 Synonyms 作者参与著作。

快速购书链接

《智能问答与深度学习》 这本书是服务于准备入门机器学习和自然语言处理的学生和软件工程师的，在理论上介绍了很多原理、算法，同时也提供很多示例程序增加实践性，这些程序被汇总到示例程序代码库，这些程序主要是帮助大家理解原理和算法的，欢迎大家下载和执行。代码库的地址是：

https://github.com/l11x0m7/book-of-qna-code

Give credits to

Word2vec by Google

Wikimedia: 训练语料来源

gensim: word2vec.py

SentenceSim: 相似度评测语料

jieba: 中文分词

License

Chunsong Public License, version 1.0

Project Sponsor

Chatopera 云服务

https://bot.chatopera.com/

Chatopera 云服务是一站式实现聊天机器人的云服务，按接口调用次数计费。Chatopera 云服务是 Chatopera 机器人平台的软件即服务实例。在云计算基础上，Chatopera 云服务属于**聊天机器人即服务**的云服务。

Chatopera 机器人平台包括知识库、多轮对话、意图识别和语音识别等组件，标准化聊天机器人开发，支持企业 OA 智能问答、HR 智能问答、智能客服和网络营销等场景。企业 IT 部门、业务部门借助 Chatopera 云服务快速让聊天机器人上线！