在 Python 中切片列表时的奇怪行为

Question

I have some data in pandas that I want to use for named entity recognition. Sample of the data is below

text
['Angie', '’s', 'is', 'my', 'favorite', 'but', 'the', 'prices', 'at', 'little', 'Tonys', 'are', 'better', '.']

tags
['B-ORG', 'I-ORG', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'B-ORG', 'I-ORG', 'O', 'O', 'O']

I ran sklearn.model_selection import train_test_split on the data

# split data
train_texts, test_texts, train_tags, test_tags = train_test_split(dataset["text"].tolist(),
                                                                dataset["tags"].tolist(),
                                                                test_size=0.20,
                                                                random_state=15)

However, when I try to print the list it gives me some weird behavior, specifically, it counts the square brackets [] and quotes '' around the text and tags as part of the test and tags. For example, when I write

print(train_texts[0][0:9], train_tags[0][0:9], sep='\n')

output
['Angie',
['B-ORG',

My question is, why is it counting the brackets and quote characters as part of the string? How can I fix it?

Answer 1

I have used DataFrame for declaration and performed the same task of splitting train_texts and test_texts and train_tags and test_tags . Kindly refer to a Solution Stated below. Then we will move ahead with the issue of [] and '' in your scenario.

# Import all the important libraries
import pandas as pd
from sklearn.model_selection import train_test_split

# Store all String data into the 'data' variable
data = {
'text' : ['Angie', '’s', 'is', 'my', 'favorite', 'but', 'the', 'prices', 'at', 'little', 'Tonys', 'are', 'better', '.'],
'tags' : ['B-ORG', 'I-ORG', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'B-ORG', 'I-ORG', 'O', 'O', 'O']}

# Store above Initialized Data into DataFrame
dataset = pd.DataFrame(data)

NOTE:- Always Print few Records of the dataset before moving ahead. Because it may happen sometimes that there was issue in your dataset which can deflect your expected result.

# Print a few records of 'dataset'
dataset

    text        tags
0   Angie       B-ORG
1   ’s          I-ORG
2   is          O
3   my          O
4   favorite    O
5   but         O
6   the         O
7   prices      O
8   at          O
9   little      B-ORG
10  Tonys       I-ORG
11  are         O
12  better      O
13  .           O

Now we can pursue the splitting part. I have used the same method which was mentioned in your question part.

# split data
train_texts, test_texts, train_tags, test_tags = train_test_split(
    dataset["text"].tolist(),
    dataset["tags"].tolist(),
    test_size=0.20,
    random_state=15)

So, after Splitting we can print a Sliced list of train_texts and train_tags

print(train_texts[0][0:9], train_tags[0][0:9], sep='\n')

Output of the above cell is stated below:-

favorite
O

As you can see, it was not printing any [] and '' in Output.

Your Question:-

Q.) Why is it counting the brackets and quote characters as part of the string? How can I fix it? A.) I don't know a proper reason behind this issue. But it may happen sometimes if your data haven't declared properly or due to any other declaration issue. But printing dataset before moving ahead is a great practice. Because you can identify the behavior of data from this method.

Solution:- Usage of DataFrame worked for me perfectly. You can use that.

Hope this Solution helps you. If you are still facing an issue kindly upload the full code. So, that we can find a solution accordingly.

Answer 2

try:

text
Angie ’s is my favorite but the prices at little Tonys are better.

tags
B-ORG I-ORG O O O O O O O B-ORG I-ORG O O O

It looks like you are trying to turn a string that is formatted to look like a list into a list. It doesn't know any better so the brackets and apostrophes are going along for the ride.