幾種簡單的文字資料預處理方法
將開頭和結尾的一些資訊去掉,使得開頭如下:
One morning, when Gregor Samsa woke from troubled dreams, he found himself transformed in his bed into a horrible vermin.
結尾如下:
And, as if in confirmation of their new dreams and good intentions, as soon as they reached their destination Grete was the first to get up and stretch out her young body.
儲存為:metamorphosis_clean.txt
載入資料:
filename='metamorphosis_clean.txt'file=open(filename,'rt')
text=file.read()
file.close()
1. 用空格分隔:
words=text.split()print(words[:100])#['One','morning,','when','Gregor','Samsa','woke','from','troubled','dreams,','he',...]
2. 用 re 分隔單詞:
和上一種方法的區別是,'armour-like' 被識別成兩個詞 'armour', 'like','What's' 變成了 'What', 's'
importre
words=re.split(r'\W+',text)
print(words[:100])
3. 用空格分隔並去掉標點:
string 裡的 string.punctuation 可以知道都有哪些算是標點符號,
maketrans() 可以建立一個空的對映表,其中 string.punctuation 是要被去掉的列表,
translate() 可以將一個字串集對映到另一個集,
也就是 'armour-like' 被識別成 'armourlike','What's' 被識別成 'Whats'
words=text.split()importstring
table=str.maketrans('','',string.punctuation)
stripped=[w.translate(table)forwinwords]
print(stripped[:100])
4. 都變成小寫:
當然大寫可以用 word.upper()。
words=[word.lower()forwordinwords]print(words[:100])
安裝 NLTK:
nltk.download() 後彈出對話方塊,選擇 all,點選 download
importnltk
nltk.download()
5. 分成句子:
用到 sent_tokenize()
fromnltkimportsent_tokenize
sentences=sent_tokenize(text)
print(sentences[0])
6. 分成單詞:
用到 word_tokenize,
這次 'armour-like' 還是 'armour-like','What's' 就是 'What', 's,
fromnltk.tokenizeimportword_tokenize
tokens=word_tokenize(text)
print(tokens[:100])
7. 過濾標點:
只保留 alphabetic,其他的濾掉,
這樣的話 “armour-like” 和 “‘s” 也被濾掉了。
fromnltk.tokenizeimportword_tokenize
tokens=word_tokenize(text)
words=[wordforwordintokensifword.isalpha()]
print(tokens[:100])
8. 過濾掉沒有深刻含義的 stop words:
在 stopwords.words('english') 可以檢視這樣的詞表。
fromnltk.corpusimportstopwords
stop_words=set(stopwords.words('english'))
words=[wforwinwordsifnotwinstop_words]
print(words[:100])
9. 轉化成詞根:
執行 porter.stem(word) 之後,單詞會變成相應的詞根形式,例如 “fishing,” “fished,” “fisher” 會變成 “fish”
fromnltk.tokenizeimportword_tokenize
tokens=word_tokenize(text)fromnltk.stem.porterimportPorterStemmer
porter=PorterStemmer()
stemmed=[porter.stem(word)forwordintokens]
print(stemmed[:100])