五、文字預處理

作者：Chris Albon

譯者：飛龍

協議：CC BY-NC-SA 4.0

詞袋

# 載入庫
import numpy as np
from sklearn.feature_extraction.text import CountVectorizer
import pandas as pd

# 建立文字
text_data = np.array(['I love Brazil. Brazil!',
                      'Sweden is best',
                      'Germany beats both'
 
])

# 建立詞袋特徵矩陣
count = CountVectorizer()
bag_of_words = count.fit_transform(text_data)

# 展示特徵矩陣
bag_of_words.toarray()

'''
array([[0, 0, 0, 2, 0, 0, 1, 0],
       [0, 1, 0, 0, 0, 1, 0, 1],
       [1, 0, 1, 0, 1, 0, 0, 0]], dtype=int64) 
'''

# 獲取特徵名稱
feature_names = count.get_feature_names()

# 檢視特徵名稱
feature_names

# ['beats', 'best', 'both', 'brazil', 'germany', 'is', 'love', 'sweden'] 
 


# 建立資料幀
pd.DataFrame(bag_of_words.toarray(), columns=feature_names)

解析 HTML

# 載入庫
from bs4 import BeautifulSoup

# 建立一些 HTML 程式碼
html = "<div class='full_name'><span style='font-weight:bold'>Masego</span> Azra</div>"

# 解析 html
soup = BeautifulSoup(html, "lxml")

# 尋找帶有 "full_name" 類的 <div>，展示文字
soup.find("div", { "class" : "full_name" }).text

# 'Masego Azra' 

移除標點

# 載入庫
import string
import numpy as np

# 建立文字
text_data = ['Hi!!!! I. Love. This. Song....', 
             '10000% Agree!!!! #LoveIT', 
             'Right?!?!']

# 建立函式，使用 string.punctuation 移除所有標點
def remove_punctuation(sentence: str) -> str:
    return sentence.translate(str.maketrans('', '', string.punctuation))

# 應用函式
[remove_punctuation(sentence) for sentence in text_data]

# ['Hi I Love This Song', '10000 Agree LoveIT', 'Right'] 

移除停止詞

# 載入庫
from nltk.corpus import stopwords

# 你第一次需要下載停止詞的集合
import nltk
nltk.download('stopwords')

'''
[nltk_data] Downloading package stopwords to
[nltk_data]     /Users/chrisalbon/nltk_data...
[nltk_data]   Package stopwords is already up-to-date!

True 
'''

# 建立單詞標記
tokenized_words = ['i', 'am', 'going', 'to', 'go', 'to', 'the', 'store', 'and', 'park']

# 載入停止詞
stop_words = stopwords.words('english')

# 展示停止詞
stop_words[:5]

# ['i', 'me', 'my', 'myself', 'we'] 

# 移除停止詞
[word for word in tokenized_words if word not in stop_words]

# ['going', 'go', 'store', 'park'] 

替換字元

# 匯入庫
import re

# 建立文字
text_data = ['Interrobang. By Aishwarya Henriette',
             'Parking And Going. By Karl Gautier',
             'Today Is The night. By Jarek Prakash']

# 移除句號
remove_periods = [string.replace('.', '') for string in text_data]

# 展示文字
remove_periods

'''
['Interrobang By Aishwarya Henriette',
 'Parking And Going By Karl Gautier',
 'Today Is The night By Jarek Prakash'] 
'''

# 建立函式
def replace_letters_with_X(string: str) -> str:
    return re.sub(r'[a-zA-Z]', 'X', string)

# 應用函式
[replace_letters_with_X(string) for string in remove_periods]

'''
['XXXXXXXXXXX XX XXXXXXXXX XXXXXXXXX',
 'XXXXXXX XXX XXXXX XX XXXX XXXXXXX',
 'XXXXX XX XXX XXXXX XX XXXXX XXXXXXX'] 
'''

詞幹提取

# 載入庫
from nltk.stem.porter import PorterStemmer

# 建立單詞標記
tokenized_words = ['i', 'am', 'humbled', 'by', 'this', 'traditional', 'meeting']

詞幹提取通過識別和刪除詞綴（例如動名詞）同時保持詞的根本意義，將詞語簡化為詞幹。 NLTK 的PorterStemmer實現了廣泛使用的 Porter 詞幹演算法。

# 建立提取器
porter = PorterStemmer()

# 應用提取器
[porter.stem(word) for word in tokenized_words]

# ['i', 'am', 'humbl', 'by', 'thi', 'tradit', 'meet'] 

移除空白

# 建立文字
text_data = ['   Interrobang. By Aishwarya Henriette     ',
             'Parking And Going. By Karl Gautier',
             '    Today Is The night. By Jarek Prakash   ']

# 移除空白
strip_whitespace = [string.strip() for string in text_data]

# 展示文字
strip_whitespace

'''
['Interrobang. By Aishwarya Henriette',
 'Parking And Going. By Karl Gautier',
 'Today Is The night. By Jarek Prakash'] 
'''

詞性標籤

# 載入庫
from nltk import pos_tag
from nltk import word_tokenize

# 建立文字
text_data = "Chris loved outdoor running"

# 使用預訓練的詞性標註器
text_tagged = pos_tag(word_tokenize(text_data))

# 展示詞性
text_tagged

# [('Chris', 'NNP'), ('loved', 'VBD'), ('outdoor', 'RP'), ('running', 'VBG')] 

輸出是一個元組列表，包含單詞和詞性的標記。 NLTK 使用 Penn Treebank 詞性標籤。

TF-IDF

# 載入庫
import numpy as np
from sklearn.feature_extraction.text import TfidfVectorizer
import pandas as pd

# 建立文字
text_data = np.array(['I love Brazil. Brazil!',
                      'Sweden is best',
                      'Germany beats both'])

# 建立 tf-idf 特徵矩陣
tfidf = TfidfVectorizer()
feature_matrix = tfidf.fit_transform(text_data)

# 展示 tf-idf 特徵矩陣
feature_matrix.toarray()

'''
array([[ 0.        ,  0.        ,  0.        ,  0.89442719,  0.        ,
         0.        ,  0.4472136 ,  0.        ],
       [ 0.        ,  0.57735027,  0.        ,  0.        ,  0.        ,
         0.57735027,  0.        ,  0.57735027],
       [ 0.57735027,  0.        ,  0.57735027,  0.        ,  0.57735027,
         0.        ,  0.        ,  0.        ]]) 
'''

# 展示 tf-idf 特徵矩陣
tfidf.get_feature_names()

# ['beats', 'best', 'both', 'brazil', 'germany', 'is', 'love', 'sweden'] 

# 建立資料幀
pd.DataFrame(feature_matrix.toarray(), columns=tfidf.get_feature_names())

文字分詞

# 載入庫
from nltk.tokenize import word_tokenize, sent_tokenize

# 建立文字
string = "The science of today is the technology of tomorrow. Tomorrow is today."

# 對文字分詞
word_tokenize(string)

'''
['The',
 'science',
 'of',
 'today',
 'is',
 'the',
 'technology',
 'of',
 'tomorrow',
 '.',
 'Tomorrow',
 'is',
 'today',
 '.'] 
'''

# 對句子分詞
sent_tokenize(string)

# ['The science of today is the technology of tomorrow.', 'Tomorrow is today.']