NLP学习（二）-语料库学习

基本语料库函数

1.古腾堡语料库

• 获取模块语料库中的文档列表

print(nltk.corpus.gutenberg.fileids()) #模块中的语料库的文档列表

• 提取特定文档的词汇，并输出文章长度

emma = nltk.corpus.gutenberg.words('austen-emma.txt')#提取文档词语
print(emma[:50])
print(len(emma))

• nltk.Text操作（详情可参考）进行初级的统计与分析

emma = nltk.Text(nltk.corpus.gutenberg.words('austen-emma.txt'))
print(emma)
print(emma.concordance("surprize"))#获取指定词汇出现的文章上下文

• 统计语料库的平均词长，平均句长，词语的多样性

from nltk.corpus import gutenberg
for fileid in gutenberg.fileids():
     num_chars = len(gutenberg.raw(fileid)) #以字母为单位进行统计
     num_words = len(gutenberg.words(fileid)) #以单词分隔符为单位进行统计
     num_sents = len(gutenberg.sents(fileid)) #以句子为单位进行统计
     num_vocab = len(set([w.lower() for w in gutenberg.words(fileid)])) #去重转化为小写在进行统计（相同词汇可能因为大小写不同而重复）
     #平均词长，平均句长， 词汇的多样性,文档名
     print(int(num_chars/num_words), int(num_words/num_sents), int(num_words/num_vocab), fileid)

• sents方法生成的是词链表，输出最长句子

macbeth_sentences = gutenberg.sents('shakespeare-macbeth.txt') #生成的是词链表
longest_len = max([len(s) for s in macbeth_sentences]) #求出最长句子长度
print([s for s in macbeth_sentences if len(s) == longest_len])#最长句子写了些什么

2.网络和聊天语料库

# #网络和聊天文本
from nltk.corpus import webtext
webtext.fileids()
for fileid in webtext.fileids():
     print(fileid, webtext.raw(fileid)[:65], '...')

3.布朗语料库（第一个百万级的电子语料库）-用于研究不同文体之间的差异

from nltk.corpus import brown
print(brown.categories()) #输出文本分类
print(brown.words(categories='news'))
print(brown.words(fileids=['cg22']))
print(brown.sents(categories=['news', 'editorial', 'reviews']))

• 查看某一分类中情态动词的词频

import nltk
from nltk.corpus import brown
news_text = brown.words(categories='news')
fdist = nltk.FreqDist([w.lower() for w in news_text])
modals = ['can', 'could', 'may', 'might', 'must', 'will']#形成想要查询的情态动词表
for m in modals:
    print(m + ':', fdist[m],)#查看情态动词的词频

• 输出不同文体之间情态动词的使用差异

import nltk
from nltk.corpus import brown
cfd = nltk.ConditionalFreqDist( #创建条件频率分布函数
     (genre, word) #文体 + 单词
     for genre in brown.categories()
     for word in brown.words(categories=genre))
genres = ['news', 'religion', 'hobbies', 'science_fiction', 'romance', 'humor']
modals = ['can', 'could', 'may', 'might', 'must', 'will']
cfd.tabulate(conditions=genres, samples=modals)#画出一个交叉表

4.路透社语料库

from nltk.corpus import reuters
print(len(reuters.fileids())) #统计文档数目
print(len(reuters.categories())) #统计文档分类数目

reuters.categories('training/9865')
reuters.categories(['training/9865', 'training/9880'])
reuters.fileids('barley') #输出分类文档

5.美国不同年份总统就职演讲语料库

from nltk.corpus import inaugural
import nltk
inaugural.fileids()
print([fileid[:4] for fileid in inaugural.fileids()])#取出文件相应的就职年份
#
cfd = nltk.ConditionalFreqDist( #特定词汇在不同就职演讲中出现的次数变化
    (target, fileid[:4])
    for fileid in inaugural.fileids()
    for w in inaugural.words(fileid)
    for target in ['america', 'citizen']
    if w.lower().startswith(target))
cfd.plot()

6.载入自己的语料库

from nltk.corpus import PlaintextCorpusReader
corpus_root = './training'  #文件目录
wordlists = PlaintextCorpusReader(corpus_root, ['pku_training.utf8','cityu_training.utf8','msr_training.utf8','pku_training.utf8'])#文件载入可以使用正则表达式进行构建文件名列表
print(wordlists.fileids())
print(wordlists.raw('pku_training.utf8'))
print(len(wordlists.words('pku_training.utf8')))
print(len(wordlists.sents('pku_training.utf8')))

7.条件频率分布，以及绘制分布图，分布表

• 创建条件概率分布

from nltk.corpus import brown
import nltk
genre_word = [(genre, word)
	for genre in ['news', 'romance']
    for word in brown.words(categories=genre)]
cfd = nltk.ConditionalFreqDist(genre_word)
print(cfd.items())

• 绘制分布图

from nltk.corpus import inaugural
import nltk
cfd = nltk.ConditionalFreqDist(
    (target, fileid[:4])
    for fileid in inaugural.fileids()
    for w in inaugural.words(fileid)
    for target in ['america', 'citizen']
    if w.lower().startswith(target))
cfd.plot()

• 绘制分布表

from nltk.corpus import udhr
import nltk
languages = ['Chickasaw', 'English', 'German_Deutsch',
    'Greenlandic_Inuktikut', 'Hungarian_Magyar', 'Ibibio_Efik']
udhr.fileids()
cfd = nltk.ConditionalFreqDist(
    (lang, len(word))
    for lang in languages
    for word in udhr.words(lang + '-Latin1'))

cfd.tabulate(conditions=['English', 'German_Deutsch'],#如果不设定就是默认都显示
    samples=range(10), cumulative=True) #限制显示样本的数目

8.使用双连词生成随机文本

import nltk

sent = ['In', 'the', 'beginning', 'God', 'created', 'the', 'heaven',
    'and', 'the', 'earth', '.']
print([i for i in nltk.bigrams(sent)])#生成的是生成器


def generate_model(cfdist, word, num=15):
    for i in range(num):
        print(word)
        word = cfdist[word].max()#下一个最大概率的连接词
text = nltk.corpus.genesis.words('english-kjv.txt')
bigrams = nltk.bigrams(text)#二连接一篇文章
# print([i for i in bigrams])
cfd = nltk.ConditionalFreqDist(bigrams)#建立条件分布图
generate_model(cfd, 'living')#生成随机文章，不断输出二连次中下一个最有可能的词语