【情感提取+情感计算+词频统计】python情感分析

import jieba.analyse
import numpy as np
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.naive_bayes import MultinomialNB
 
# 加载数据
def load_data(pos_file, neg_file):
    with open(pos_file, 'r', encoding='utf-8') as f:
        pos_data = f.readlines()
    with open(neg_file, 'r', encoding='utf-8') as f:
        neg_data = f.readlines()
    return pos_data, neg_data
 
# 切词并生成词频统计
def generate_word_freq(data, cut_all=False):
    # 使用结巴分词
    words = []
    for sentence in data:
        words.extend(jieba.cut(sentence, cut_all=cut_all))
    
    # 使用TfidfVectorizer计算词频
    vectorizer = TfidfVectorizer(stop_words='english')
    X = vectorizer.fit_transform([word for word in words if word not in vectorizer.vocabulary_]).toarray()
    
    # 返回词频矩阵和词汇表
    return X, vectorizer.vocabulary_
 
# 训练情感判别模型
def train_model(pos_data, neg_data):
    pos_data, neg_data = load_data(pos_file, neg_file)
    pos_X, pos_vocab = generate_word_freq(pos_data)
    neg_X, neg_vocab = generate_word_freq(neg_data)
    
    # 合并词汇表并重新索引词频矩阵
    vocab = pos_vocab.copy()
    neg_vocab_idx = {word: idx for idx, word in enumerate(neg_vocab, len(pos_vocab))}
    vocab.update(neg_vocab_idx)
    X = np.concatenate([pos_X, neg_X], axis=0)
    y = np.concatenate([np.ones(len(pos_X)), np.zeros(len(neg_X))])
    
    # 训练多项朴布斯分类器
    vectorizer = TfidfVectorizer(vocabulary=vocab)
    X_train = vectorizer.fit_transform(data).toarray()
    classifier = MultinomialNB()
    classifier.fit(X_train, y)
    return classifier, vectorizer
 
# 主程序
if __name__ == '__main__':
    pos_file = 'pos.txt'  # 正面评论数据文件
    neg_file = 'neg.txt'  # 负面评论数据文件
    classifier, vectorizer = train_model(pos_file, neg_file)
 
    # 示例：评估一条新的评论情感
    sentence = '这家酒店位置很好，环境清洁'
    X_test = vectorizer.transform([sentence]).toarray()
    y_pred = classifier.predict(X_test)
    print(f'情感: {"正面" if y_pred else "负面"}')

这段代码首先加载了正面和负面的评论数据，然后使用结巴分词库进行了中文分词，并使用TfidfVectorizer计算了词