不可散列的类型：停用词的“列表”错误

Question

这是我的代码

CSV 文件的网址：https://github.com/eugeneketeni/web-mining-final-project/blob/master/Test_file.csv

import pandas as pd

data = pd.read_csv("https://raw.githubusercontent.com/eugeneketeni/web- 
mining-final-project/master/Test_file.csv")

import nltk
from nltk import word_tokenize, sent_tokenize


data['text'] = data.loc[:, 'text'].astype(str)

text = data.loc[:, "text"].astype(str)
tokenizer = [word_tokenize(text[i]) for i in range(len(text))]
print(tokenizer)

filtered_sentence = []


from nltk.corpus import stopwords
stopwords = set(stopwords.words('english'))

filtered_sentence = [w for w in tokenizer if not w in stopwords]
print(filtered_sentence) 

我的标记器可以工作，但是当我尝试删除默认停用词时，我不断收到“unhashable type: 'list'”错误。我不确定到底发生了什么。我将不胜感激任何帮助。谢谢。

Answer 1

TL;DR

from nltk import word_tokenize
from nltk.corpus import stopwords

import pandas as pd

stoplist = set(stopwords.words('english'))

data = pd.read_csv("Test_file.csv")

data['filtered_text'] = data['text'].astype(str).apply(lambda line: [token for token in word_tokenize(line) if token not in stoplist])

长期

请参阅Why is my NLTK function slow when processing the DataFrame?了解更多详细说明：

• 标记数据框中的文本
• 删除停用词
• 其他相关清洁工艺

---

为了更好，推特文本处理

pip3 install -U nltk[twitter]

然后使用这个：

从 nltk.corpus 导入停用词

from nltk.tokenize import TweetTokenizer

import pandas as pd

word_tokenize = TweetTokenizer().tokenize

stoplist = set(stopwords.words('english'))

data = pd.read_csv("Test_file.csv")

data['filtered_text'] = data['text'].astype(str).apply(lambda line: [token for token in word_tokenize(line) if token not in stoplist])