基于朴素贝叶斯实现垃圾邮件分类，不用sklearn

可以使用Python中的nltk库来实现基于朴素贝叶斯的垃圾邮件分类，以下是一个简单的实现代码：

import nltk
import random
from nltk.corpus import stopwords
from nltk.tokenize import word_tokenize

# 读取垃圾邮件和正常邮件的数据
spam = open('spam.txt', 'r').read()
ham = open('ham.txt', 'r').read()

# 分词和去除停用词
spam_words = word_tokenize(spam)
ham_words = word_tokenize(ham)
stop_words = set(stopwords.words('english'))

spam_words = [word.lower() for word in spam_words if word.isalpha() and word.lower() not in stop_words]
ham_words = [word.lower() for word in ham_words if word.isalpha() and word.lower() not in stop_words]

# 构建词汇表
all_words = set(spam_words + ham_words)
word_features = list(all_words)

# 构建训练集和测试集
def extract_features(document):
    document_words = set(document)
    features = {}
    for word in word_features:
        features[word] = (word in document_words)
    return features

spam_features = [(extract_features(spam_words), 'spam') for spam_words in spam]
ham_features = [(extract_features(ham_words), 'ham') for ham_words in ham]

train_set = spam_features + ham_features
random.shuffle(train_set)

# 训练模型
classifier = nltk.NaiveBayesClassifier.train(train_set)

# 测试模型
test_spam = open('test_spam.txt', 'r').read()
test_ham = open('test_ham.txt', 'r').read()

test_spam_words = word_tokenize(test_spam)
test_ham_words = word_tokenize(test_ham)

test_spam_features = extract_features(test_spam_words)
test_ham_features = extract_features(test_ham_words)

print('Test Spam:', classifier.classify(test_spam_features))
print('Test Ham:', classifier.classify(test_ham_features))

其中，spam.txt和ham.txt分别是垃圾邮件和正常邮件的数据，test_spam.txt和test_ham.txt是用于测试的数据。