导入所需库

import jieba import pandas as pd from sklearn.feature_extraction.text import CountVectorizer, TfidfVectorizer from gensim.models import Word2Vec

读取数据集

corpus = pd.read_csv('corpus.txt', sep=' ', header=None, names=['label', 'text'])

分词处理

corpus['text_cut'] = corpus['text'].apply(lambda x: ' '.join(jieba.cut(x)))

停用词处理

stopwords = pd.read_csv('stopwords.txt', sep=' ', header=None, names=['stopword'], encoding='utf-8') corpus['text_cut_stop'] = corpus['text_cut'].apply(lambda x: ' '.join([word for word in x.split() if word not in stopwords['stopword'].tolist()]) )

文本向量化

one-hot编码

cv = CountVectorizer() one_hot = cv.fit_transform(corpus['text_cut_stop']).toarray()

TF-IDF编码

tfidf = TfidfVectorizer() tf_idf = tfidf.fit_transform(corpus['text_cut_stop']).toarray()

Word2Vec编码

w2v = Word2Vec([text.split() for text in corpus['text_cut_stop']], size=100, min_count=1) w2v_vec = [] for text in corpus['text_cut_stop']: vec = [] for word in text.split(): vec.append(w2v.wv[word]) w2v_vec.append(sum(vec) / len(vec))