特征构造

1import pandas as pd
2
3df = pd.read_csv("reviews.csv")
4df['review_length'] = df['review_text'].apply(lambda x: len(x.split()))
5author_review_count = df['author_name'].value_counts().to_dict()
6df['author_review_count'] = df['author_name'].map(author_review_count)
7print(df[['review_length', 'author_review_count']])

1import pandas as pd
2import spacy
3
4df = pd.read_csv("reviews.csv")
5nlp = spacy.load('en_core_web_sm')
6def extract_named_entities(text):
7    doc = nlp(text)
8    named_entities = [ent.text for ent in doc.ents]
9    return named_entities
10df['named_entities'] = df['review_text'].apply(extract_named_entities)
11print(df['named_entities'])