“结巴”中文分词:做最好的 Python 中文分词组件
"Jieba" (Chinese for "to stutter") Chinese text segmentation: built to be the best Python Chinese word segmentation module.
- Scroll down for English documentation.
-
支持三种分词模式:
- 精确模式,试图将句子最精确地切开,适合文本分析;
- 全模式,把句子中所有的可以成词的词语都扫描出来, 速度非常快,但是不能解决歧义;
- 搜索引擎模式,在精确模式的基础上,对长词再次切分,提高召回率,适合用于搜索引擎分词。
-
支持繁体分词
-
支持自定义词典
-
MIT 授权协议
http://jiebademo.ap01.aws.af.cm/
(Powered by Appfog)
网站代码:https://github.com/fxsjy/jiebademo
代码对 Python 2/3 均兼容
- 全自动安装:
easy_install jieba
或者pip install jieba
/pip3 install jieba
- 半自动安装:先下载 http://pypi.python.org/pypi/jieba/ ,解压后运行
python setup.py install
- 手动安装:将 jieba 目录放置于当前目录或者 site-packages 目录
- 通过
import jieba
来引用
- 基于前缀词典实现高效的词图扫描,生成句子中汉字所有可能成词情况所构成的有向无环图 (DAG)
- 采用了动态规划查找最大概率路径, 找出基于词频的最大切分组合
- 对于未登录词,采用了基于汉字成词能力的 HMM 模型,使用了 Viterbi 算法
- :分词
jieba.cut
方法接受三个输入参数: 需要分词的字符串;cut_all 参数用来控制是否采用全模式;HMM 参数用来控制是否使用 HMM 模型jieba.cut_for_search
方法接受两个参数:需要分词的字符串;是否使用 HMM 模型。该方法适合用于搜索引擎构建倒排索引的分词,粒度比较细- 待分词的字符串可以是 unicode 或 UTF-8 字符串、GBK 字符串。注意:不建议直接输入 GBK 字符串,可能无法预料地错误解码成 UTF-8
jieba.cut
以及jieba.cut_for_search
返回的结构都是一个可迭代的 generator,可以使用 for 循环来获得分词后得到的每一个词语(unicode),也可以用 list(jieba.cut(...)) 转化为 list
代码示例( 分词 )
#encoding=utf-8
import jieba
seg_list = jieba.cut("我来到北京清华大学", cut_all=True)
print("Full Mode: " + "/ ".join(seg_list)) # 全模式
seg_list = jieba.cut("我来到北京清华大学", cut_all=False)
print("Default Mode: " + "/ ".join(seg_list)) # 精确模式
seg_list = jieba.cut("他来到了网易杭研大厦") # 默认是精确模式
print(", ".join(seg_list))
seg_list = jieba.cut_for_search("小明硕士毕业于中国科学院计算所,后在日本京都大学深造") # 搜索引擎模式
print(", ".join(seg_list))
输出:
【全模式】: 我/ 来到/ 北京/ 清华/ 清华大学/ 华大/ 大学
【精确模式】: 我/ 来到/ 北京/ 清华大学
【新词识别】:他, 来到, 了, 网易, 杭研, 大厦 (此处,“杭研”并没有在词典中,但是也被Viterbi算法识别出来了)
【搜索引擎模式】: 小明, 硕士, 毕业, 于, 中国, 科学, 学院, 科学院, 中国科学院, 计算, 计算所, 后, 在, 日本, 京都, 大学, 日本京都大学, 深造
- :添加自定义词典
-
开发者可以指定自己自定义的词典,以便包含 jieba 词库里没有的词。虽然 jieba 有新词识别能力,但是自行添加新词可以保证更高的正确率
-
用法: jieba.load_userdict(file_name) # file_name 为自定义词典的路径
-
词典格式和
dict.txt
一样,一个词占一行;每一行分三部分,一部分为词语,另一部分为词频(可省略),最后为词性(可省略),用空格隔开 -
词频可省略,使用计算出的能保证分出该词的词频
-
范例:
-
自定义词典:https://github.com/fxsjy/jieba/blob/master/test/userdict.txt
-
用法示例:https://github.com/fxsjy/jieba/blob/master/test/test_userdict.py
-
之前: 李小福 / 是 / 创新 / 办 / 主任 / 也 / 是 / 云 / 计算 / 方面 / 的 / 专家 /
-
加载自定义词库后: 李小福 / 是 / 创新办 / 主任 / 也 / 是 / 云计算 / 方面 / 的 / 专家 /
-
-
-
使用
add_word(word, freq=None, tag=None)
和del_word(word)
可在程序中动态修改词典。 -
使用
suggest_freq(segment, tune=True)
可调节单个词语的词频,使其能(或不能)被分出来。 -
注意:自动计算的词频在使用 HMM 新词发现功能时可能无效。
代码示例:
>>> print('/'.join(jieba.cut('如果放到post中将出错。', HMM=False)))
如果/放到/post/中将/出错/。
>>> jieba.suggest_freq(('中', '将'), True)
494
>>> print('/'.join(jieba.cut('如果放到post中将出错。', HMM=False)))
如果/放到/post/中/将/出错/。
>>> print('/'.join(jieba.cut('「台中」正确应该不会被切开', HMM=False)))
「/台/中/」/正确/应该/不会/被/切开
>>> jieba.suggest_freq('台中', True)
69
>>> print('/'.join(jieba.cut('「台中」正确应该不会被切开', HMM=False)))
「/台中/」/正确/应该/不会/被/切开
- "通过用户自定义词典来增强歧义纠错能力" --- fxsjy#14
- :关键词提取
- jieba.analyse.extract_tags(sentence,topK,withWeight) #需要先
import jieba.analyse
- sentence 为待提取的文本
- topK 为返回几个 TF/IDF 权重最大的关键词,默认值为 20
- withWeight 为是否一并返回关键词权重值,默认值为 False
代码示例 (关键词提取)
https://github.com/fxsjy/jieba/blob/master/test/extract_tags.py
关键词提取所使用逆向文件频率(IDF)文本语料库可以切换成自定义语料库的路径
- 用法: jieba.analyse.set_idf_path(file_name) # file_name为自定义语料库的路径
- 自定义语料库示例:https://github.com/fxsjy/jieba/blob/master/extra_dict/idf.txt.big
- 用法示例:https://github.com/fxsjy/jieba/blob/master/test/extract_tags_idfpath.py
关键词提取所使用停止词(Stop Words)文本语料库可以切换成自定义语料库的路径
- 用法: jieba.analyse.set_stop_words(file_name) # file_name为自定义语料库的路径
- 自定义语料库示例:https://github.com/fxsjy/jieba/blob/master/extra_dict/stop_words.txt
- 用法示例:https://github.com/fxsjy/jieba/blob/master/test/extract_tags_stop_words.py
关键词一并返回关键词权重值示例
算法论文: TextRank: Bringing Order into Texts
- 将待抽取关键词的文本进行分词
- 以固定窗口大小(我选的5,可适当调整),词之间的共现关系,构建图
- 计算图中节点的PageRank,注意是无向带权图
jieba.analyse.textrank(raw_text)
来自__main__
的示例结果:
吉林 1.0
欧亚 0.864834432786
置业 0.553465925497
实现 0.520660869531
收入 0.379699688954
增资 0.355086023683
子公司 0.349758490263
全资 0.308537396283
城市 0.306103738053
商业 0.304837414946
- : 词性标注
- 标注句子分词后每个词的词性,采用和 ictclas 兼容的标记法
- 用法示例
>>> import jieba.posseg as pseg
>>> words = pseg.cut("我爱北京天安门")
>>> for w in words:
... print('%s %s' % (w.word, w.flag))
...
我 r
爱 v
北京 ns
天安门 ns
- : 并行分词
-
原理:将目标文本按行分隔后,把各行文本分配到多个 python 进程并行分词,然后归并结果,从而获得分词速度的可观提升
-
基于 python 自带的 multiprocessing 模块,目前暂不支持 windows
-
用法:
jieba.enable_parallel(4)
# 开启并行分词模式,参数为并行进程数jieba.disable_parallel()
# 关闭并行分词模式
-
例子:https://github.com/fxsjy/jieba/blob/master/test/parallel/test_file.py
-
实验结果:在 4 核 3.4GHz Linux 机器上,对金庸全集进行精确分词,获得了 1MB/s 的速度,是单进程版的 3.3 倍。
- : Tokenize:返回词语在原文的起始位置
- 注意,输入参数只接受 unicode
- 默认模式
result = jieba.tokenize(u'永和服装饰品有限公司')
for tk in result:
print("word %s\t\t start: %d \t\t end:%d" % (tk[0],tk[1],tk[2]))
word 永和 start: 0 end:2
word 服装 start: 2 end:4
word 饰品 start: 4 end:6
word 有限公司 start: 6 end:10
- 搜索模式
result = jieba.tokenize(u'永和服装饰品有限公司',mode='search')
for tk in result:
print("word %s\t\t start: %d \t\t end:%d" % (tk[0],tk[1],tk[2]))
word 永和 start: 0 end:2
word 服装 start: 2 end:4
word 饰品 start: 4 end:6
word 有限 start: 6 end:8
word 公司 start: 8 end:10
word 有限公司 start: 6 end:10
- : ChineseAnalyzer for Whoosh 搜索引擎
- 引用:
from jieba.analyse import ChineseAnalyzer
- 用法示例:https://github.com/fxsjy/jieba/blob/master/test/test_whoosh.py
- : 命令行分词
使用示例:cat news.txt | python -m jieba > cut_result.txt
命令行选项(翻译):
使用: python -m jieba [options] filename
结巴命令行界面。
固定参数:
filename 输入文件
可选参数:
-h, --help 显示此帮助信息并退出
-d [DELIM], --delimiter [DELIM]
使用 DELIM 分隔词语,而不是用默认的' / '。
若不指定 DELIM,则使用一个空格分隔。
-D DICT, --dict DICT 使用 DICT 代替默认词典
-u USER_DICT, --user-dict USER_DICT
使用 USER_DICT 作为附加词典,与默认词典或自定义词典配合使用
-a, --cut-all 全模式分词
-n, --no-hmm 不使用隐含马尔可夫模型
-q, --quiet 不输出载入信息到 STDERR
-V, --version 显示版本信息并退出
如果没有指定文件名,则使用标准输入。
--help
选项输出:
$> python -m jieba --help
usage: python -m jieba [options] filename
Jieba command line interface.
positional arguments:
filename input file
optional arguments:
-h, --help show this help message and exit
-d [DELIM], --delimiter [DELIM]
use DELIM instead of ' / ' for word delimiter; or a
space if it is used without DELIM
-D DICT, --dict DICT use DICT as dictionary
-u USER_DICT, --user-dict USER_DICT
use USER_DICT together with the default dictionary or
DICT (if specified)
-a, --cut-all full pattern cutting
-n, --no-hmm don't use the Hidden Markov Model
-q, --quiet don't print loading messages to stderr
-V, --version show program's version number and exit
If no filename specified, use STDIN instead.
jieba 采用延迟加载,"import jieba" 不会立即触发词典的加载,一旦有必要才开始加载词典构建前缀字典。如果你想手工初始 jieba,也可以手动初始化。
import jieba
jieba.initialize() # 手动初始化(可选)
在 0.28 之前的版本是不能指定主词典的路径的,有了延迟加载机制后,你可以改变主词典的路径:
jieba.set_dictionary('data/dict.txt.big')
例子: https://github.com/fxsjy/jieba/blob/master/test/test_change_dictpath.py
-
占用内存较小的词典文件 https://github.com/fxsjy/jieba/raw/master/extra_dict/dict.txt.small
-
支持繁体分词更好的词典文件 https://github.com/fxsjy/jieba/raw/master/extra_dict/dict.txt.big
下载你所需要的词典,然后覆盖 jieba/dict.txt 即可;或者用 jieba.set_dictionary('data/dict.txt.big')
作者:piaolingxue 地址:https://github.com/huaban/jieba-analysis
作者:yanyiwu 地址:https://github.com/yanyiwu/cppjieba
作者:yanyiwu 地址:https://github.com/yanyiwu/nodejieba
作者:falood 地址:https://github.com/falood/exjieba
作者:qinwf 地址:https://github.com/qinwf/jiebaR
作者:yanyiwu 地址:https://github.com/yanyiwu/iosjieba
- 1.5 MB / Second in Full Mode
- 400 KB / Second in Default Mode
- 测试环境: Intel(R) Core(TM) i7-2600 CPU @ 3.4GHz;《围城》.txt
详见: fxsjy#7
P(台中) < P(台)×P(中),“台中”词频不够导致其成词概率较低
解决方法:强制调高词频
jieba.add_word('台中')
或者 jieba.suggest_freq('台中', True)
解决方法:强制调低词频
jieba.suggest_freq(('今天', '天气'), True)
或者直接删除该词 jieba.del_word('今天天气')
解决方法:关闭新词发现
jieba.cut('丰田太省了', HMM=False)
jieba.cut('我们中出了一个叛徒', HMM=False)
更多问题请点击:https://github.com/fxsjy/jieba/issues?sort=updated&state=closed
https://github.com/fxsjy/jieba/blob/master/Changelog
"Jieba" (Chinese for "to stutter") Chinese text segmentation: built to be the best Python Chinese word segmentation module.
- Support three types of segmentation mode:
- Accurate Mode attempts to cut the sentence into the most accurate segmentations, which is suitable for text analysis.
- Full Mode gets all the possible words from the sentence. Fast but not accurate.
- Search Engine Mode, based on the Accurate Mode, attempts to cut long words into several short words, which can raise the recall rate. Suitable for search engines.
- Supports Traditional Chinese
- Supports customized dictionaries
- MIT License
http://jiebademo.ap01.aws.af.cm/
(Powered by Appfog)
- Fully automatic installation:
easy_install jieba
orpip install jieba
- Semi-automatic installation: Download http://pypi.python.org/pypi/jieba/ , run
python setup.py install
after extracting. - Manual installation: place the
jieba
directory in the current directory or pythonsite-packages
directory. import jieba
.
- Based on a prefix dictionary structure to achieve efficient word graph scanning. Build a directed acyclic graph (DAG) for all possible word combinations.
- Use dynamic programming to find the most probable combination based on the word frequency.
- For unknown words, a HMM-based model is used with the Viterbi algorithm.
- : Cut
- The
jieba.cut
function accepts three input parameters: the first parameter is the string to be cut; the second parameter iscut_all
, controlling the cut mode; the third parameter is to control whether to use the Hidden Markov Model. jieba.cut_for_search
accepts two parameter: the string to be cut; whether to use the Hidden Markov Model. This will cut the sentence into short words suitable for search engines.- The input string can be an unicode/str object, or a str/bytes object which is encoded in UTF-8 or GBK. Note that using GBK encoding is not recommended because it may be unexpectly decoded as UTF-8.
jieba.cut
andjieba.cut_for_search
returns an generator, from which you can use afor
loop to get the segmentation result (in unicode), orlist(jieba.cut( ... ))
to create a list.
Code example: segmentation
#encoding=utf-8
import jieba
seg_list = jieba.cut("我来到北京清华大学", cut_all=True)
print("Full Mode: " + "/ ".join(seg_list)) # 全模式
seg_list = jieba.cut("我来到北京清华大学", cut_all=False)
print("Default Mode: " + "/ ".join(seg_list)) # 默认模式
seg_list = jieba.cut("他来到了网易杭研大厦")
print(", ".join(seg_list))
seg_list = jieba.cut_for_search("小明硕士毕业于中国科学院计算所,后在日本京都大学深造") # 搜索引擎模式
print(", ".join(seg_list))
Output:
[Full Mode]: 我/ 来到/ 北京/ 清华/ 清华大学/ 华大/ 大学
[Accurate Mode]: 我/ 来到/ 北京/ 清华大学
[Unknown Words Recognize] 他, 来到, 了, 网易, 杭研, 大厦 (In this case, "杭研" is not in the dictionary, but is identified by the Viterbi algorithm)
[Search Engine Mode]: 小明, 硕士, 毕业, 于, 中国, 科学, 学院, 科学院, 中国科学院, 计算, 计算所, 后, 在, 日本, 京都, 大学, 日本京都大学, 深造
- : Add a custom dictionary
### Load dictionary
-
Developers can specify their own custom dictionary to be included in the jieba default dictionary. Jieba is able to identify new words, but adding your own new words can ensure a higher accuracy.
-
Usage:
jieba.load_userdict(file_name) # file_name is the path of the custom dictionary
-
The dictionary format is the same as that of
analyse/idf.txt
: one word per line; each line is divided into two parts, the first is the word itself, the other is the word frequency, separated by a space -
Example:
云计算 5 李小福 2 创新办 3 [Before]: 李小福 / 是 / 创新 / 办 / 主任 / 也 / 是 / 云 / 计算 / 方面 / 的 / 专家 / [After]: 李小福 / 是 / 创新办 / 主任 / 也 / 是 / 云计算 / 方面 / 的 / 专家 /
-
Use
add_word(word, freq=None, tag=None)
anddel_word(word)
to modify the dictionary dynamically in programs. -
Use
suggest_freq(segment, tune=True)
to adjust the frequency of a single word so that it can (or cannot) be segmented. -
Note that HMM may affect the final result.
Example:
>>> print('/'.join(jieba.cut('如果放到post中将出错。', HMM=False)))
如果/放到/post/中将/出错/。
>>> jieba.suggest_freq(('中', '将'), True)
494
>>> print('/'.join(jieba.cut('如果放到post中将出错。', HMM=False)))
如果/放到/post/中/将/出错/。
>>> print('/'.join(jieba.cut('「台中」正确应该不会被切开', HMM=False)))
「/台/中/」/正确/应该/不会/被/切开
>>> jieba.suggest_freq('台中', True)
69
>>> print('/'.join(jieba.cut('「台中」正确应该不会被切开', HMM=False)))
「/台中/」/正确/应该/不会/被/切开
- : Keyword Extraction
jieba.analyse.extract_tags(sentence,topK,withWeight) # needs to first import jieba.analyse
sentence
: the text to be extractedtopK
: return how many keywords with the highest TF/IDF weights. The default value is 20withWeight
: whether return TF/IDF weights with the keywords. The default value is False
Example (keyword extraction)
https://github.com/fxsjy/jieba/blob/master/test/extract_tags.py
Developers can specify their own custom IDF corpus in jieba keyword extraction
- Usage:
jieba.analyse.set_idf_path(file_name) # file_name is the path for the custom corpus
- Custom Corpus Sample:https://github.com/fxsjy/jieba/blob/master/extra_dict/idf.txt.big
- Sample Code:https://github.com/fxsjy/jieba/blob/master/test/extract_tags_idfpath.py
Developers can specify their own custom stop words corpus in jieba keyword extraction
- Usage:
jieba.analyse.set_stop_words(file_name) # file_name is the path for the custom corpus
- Custom Corpus Sample:https://github.com/fxsjy/jieba/blob/master/extra_dict/stop_words.txt
- Sample Code:https://github.com/fxsjy/jieba/blob/master/test/extract_tags_stop_words.py
There's also a TextRank implementation available.
Use: jieba.analyse.textrank(raw_text)
.
- : Part of Speech Tagging
- Tags the POS of each word after segmentation, using labels compatible with ictclas.
- Example:
>>> import jieba.posseg as pseg
>>> words = pseg.cut("我爱北京天安门")
>>> for w in words:
... print('%s %s' % (w.word, w.flag))
...
我 r
爱 v
北京 ns
天安门 ns
- : Parallel Processing
-
Principle: Split target text by line, assign the lines into multiple Python processes, and then merge the results, which is considerably faster.
-
Based on the multiprocessing module of Python.
-
Usage:
jieba.enable_parallel(4)
# Enable parallel processing. The parameter is the number of processes.jieba.disable_parallel()
# Disable parallel processing.
-
Example: https://github.com/fxsjy/jieba/blob/master/test/parallel/test_file.py
-
Result: On a four-core 3.4GHz Linux machine, do accurate word segmentation on Complete Works of Jin Yong, and the speed reaches 1MB/s, which is 3.3 times faster than the single-process version.
- : Tokenize: return words with position
- The input must be unicode
- Default mode
result = jieba.tokenize(u'永和服装饰品有限公司')
for tk in result:
print("word %s\t\t start: %d \t\t end:%d" % (tk[0],tk[1],tk[2]))
word 永和 start: 0 end:2
word 服装 start: 2 end:4
word 饰品 start: 4 end:6
word 有限公司 start: 6 end:10
- Search mode
result = jieba.tokenize(u'永和服装饰品有限公司',mode='search')
for tk in result:
print("word %s\t\t start: %d \t\t end:%d" % (tk[0],tk[1],tk[2]))
word 永和 start: 0 end:2
word 服装 start: 2 end:4
word 饰品 start: 4 end:6
word 有限 start: 6 end:8
word 公司 start: 8 end:10
word 有限公司 start: 6 end:10
- : ChineseAnalyzer for Whoosh
from jieba.analyse import ChineseAnalyzer
- Example: https://github.com/fxsjy/jieba/blob/master/test/test_whoosh.py
- : Command Line Interface
$> python -m jieba --help
usage: python -m jieba [options] filename
Jieba command line interface.
positional arguments:
filename input file
optional arguments:
-h, --help show this help message and exit
-d [DELIM], --delimiter [DELIM]
use DELIM instead of ' / ' for word delimiter; or a
space if it is used without DELIM
-D DICT, --dict DICT use DICT as dictionary
-u USER_DICT, --user-dict USER_DICT
use USER_DICT together with the default dictionary or
DICT (if specified)
-a, --cut-all full pattern cutting
-n, --no-hmm don't use the Hidden Markov Model
-q, --quiet don't print loading messages to stderr
-V, --version show program's version number and exit
If no filename specified, use STDIN instead.
By default, Jieba don't build the prefix dictionary unless it's necessary. This takes 1-3 seconds, after which it is not initialized again. If you want to initialize Jieba manually, you can call:
import jieba
jieba.initialize() # (optional)
You can also specify the dictionary (not supported before version 0.28) :
jieba.set_dictionary('data/dict.txt.big')
It is possible to use your own dictionary with Jieba, and there are also two dictionaries ready for download:
-
A smaller dictionary for a smaller memory footprint: https://github.com/fxsjy/jieba/raw/master/extra_dict/dict.txt.small
-
There is also a bigger dictionary that has better support for traditional Chinese (繁體): https://github.com/fxsjy/jieba/raw/master/extra_dict/dict.txt.big
By default, an in-between dictionary is used, called dict.txt
and included in the distribution.
In either case, download the file you want, and then call jieba.set_dictionary('data/dict.txt.big')
or just replace the existing dict.txt
.
- 1.5 MB / Second in Full Mode
- 400 KB / Second in Default Mode
- Test Env: Intel(R) Core(TM) i7-2600 CPU @ 3.4GHz;《围城》.txt