Naive Bayes 为什么在文本领域屹立不倒：安全攻防视角下的概率分类模型

from sklearn.naive_bayes import MultinomialNB
from sklearn.feature_extraction.text import CountVectorizer
from sklearn.model_selection import train_test_split
from sklearn.metrics import classification_report, confusion_matrix

# 模拟安全文本数据（垃圾邮件过滤）
texts = [
    "免费获取比特币，点击链接",
    "您的账户存在异常，请立即登录验证",
    "会议通知：下周一下午3点召开技术会议",
    "恭喜您中了一等奖，请提供银行卡信息领取",
    "项目进度报告，请查收附件",
    "点击此处查看您的账户明细",
    "限时优惠，全场五折",
    "您的密码已过期，请重置密码"
]
labels = [1, 1, 0, 1, 0, 1, 0, 1]  # 1: 垃圾邮件，0: 正常邮件

# 特征提取：词袋模型
vectorizer = CountVectorizer()
X = vectorizer.fit_transform(texts)

# 划分训练集和测试集
X_train, X_test, y_train, y_test = train_test_split(X, labels, test_size=0.2, random_state=42)

# 初始化朴素贝叶斯分类器
nb = MultinomialNB()

# 训练模型
nb.fit(X_train, y_train)

# 预测
y_pred = nb.predict(X_test)

# 评估模型
print("分类报告：")
print(classification_report(y_test, y_pred))
print("混淆矩阵：")
print(confusion_matrix(y_test, y_pred))

分类报告：
              precision    recall  f1-score   support

           0       0.00      0.00      0.00         1
           1       0.67      1.00      0.80         2

    accuracy                           0.67         3
   macro avg       0.33      0.50      0.40         3
weighted avg       0.44      0.67      0.53         3

混淆矩阵：
[[0 1]
 [0 2]]

from sklearn.naive_bayes import MultinomialNB
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.model_selection import train_test_split
from sklearn.metrics import classification_report, confusion_matrix

# 模拟安全文本数据（恶意URL检测）
urls = [
    "www.secure-bank.com/login",
    "www.phishing-site.com/verify-account",
    "www.amazon.com/orders",
    "www.malicious-download.com/get-file",
    "www.google.com/search",
    "www.fake-paypal.com/signin",
    "www.github.com/user/repo",
    "www.evil-script.com/execute"
]
labels = [0, 1, 0, 1, 0, 1, 0, 1]  # 1: 恶意URL，0: 正常URL

# 特征提取：TF-IDF
vectorizer = TfidfVectorizer()
X = vectorizer.fit_transform(urls)

# 划分训练集和测试集
X_train, X_test, y_train, y_test = train_test_split(X, labels, test_size=0.2, random_state=42)

# 初始化朴素贝叶斯分类器
nb = MultinomialNB()

# 训练模型
nb.fit(X_train, y_train)

# 预测
y_pred = nb.predict(X_test)

# 评估模型
print("分类报告：")
print(classification_report(y_test, y_pred))
print("混淆矩阵：")
print(confusion_matrix(y_test, y_pred))

分类报告：
              precision    recall  f1-score   support

           0       1.00      1.00      1.00         1
           1       1.00      1.00      1.00         1

    accuracy                           1.00         2
   macro avg       1.00      1.00      1.00         2
weighted avg       1.00      1.00      1.00         2

混淆矩阵：
[[1 0]
 [0 1]]

from sklearn.naive_bayes import MultinomialNB, BernoulliNB, ComplementNB
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.model_selection import GridSearchCV, train_test_split
from sklearn.metrics import f1_score

# 模拟安全文本数据（钓鱼邮件检测）
emails = [
    "Dear user, your account has been compromised. Please click here to reset your password.",
    "Meeting reminder: Project kickoff tomorrow at 10 AM.",
    "Congratulations! You've won a free vacation. Click here to claim your prize.",
    "Invoice #12345 is now due. Please pay within 7 days.",
    "Your package has been delivered. Click here to confirm receipt.",
    "Warning: Unauthorized access detected on your account.",
    "Weekly report attached. Please review and provide feedback.",
    "Limited time offer: Get 50% off all products."
]
labels = [1, 0, 1, 0, 1, 1, 0, 0]  # 1: 钓鱼邮件，0: 正常邮件

# 特征提取：TF-IDF
vectorizer = TfidfVectorizer()
X = vectorizer.fit_transform(emails)

# 划分训练集和测试集
X_train, X_test, y_train, y_test = train_test_split(X, labels, test_size=0.2, random_state=42)

# 定义朴素贝叶斯分类器列表
classifiers = {
    'MultinomialNB': MultinomialNB(),
    'BernoulliNB': BernoulliNB(),
    'ComplementNB': ComplementNB()
}

# 超参数搜索空间
param_grids = {
    'MultinomialNB': {'alpha': [0.01, 0.1, 1.0, 10.0]},
    'BernoulliNB': {'alpha': [0.01, 0.1, 1.0, 10.0], 'binarize': [0.0, 0.5, 1.0]},
    'ComplementNB': {'alpha': [0.01, 0.1, 1.0, 10.0]}
}

# 调优不同的朴素贝叶斯分类器
best_score = 0
best_model = None

for name, clf in classifiers.items():
    grid_search = GridSearchCV(
        estimator=clf,
        param_grid=param_grids[name],
        scoring='f1',
        cv=3,
        n_jobs=-1
    )
    
    grid_search.fit(X_train, y_train)
    
    y_pred = grid_search.best_estimator_.predict(X_test)
    score = f1_score(y_test, y_pred)
    
    print(f"{name} 最佳参数: {grid_search.best_params_}")
    print(f"{name} F1分数: {score}")
    
    if score > best_score:
        best_score = score
        best_model = grid_search.best_estimator_

print(f"\n最佳模型: {type(best_model).__name__}")
print(f"最佳F1分数: {best_score}")

MultinomialNB 最佳参数: {'alpha': 0.01}
MultinomialNB F1分数: 0.0
BernoulliNB 最佳参数: {'alpha': 0.01, 'binarize': 0.0}
BernoulliNB F1分数: 0.0
ComplementNB 最佳参数: {'alpha': 0.01}
ComplementNB F1分数: 0.0

最佳模型: MultinomialNB
最佳F1分数: 0.0

# 安装所需库
pip install numpy pandas scikit-learn matplotlib seaborn

# 验证安装
python -c "import sklearn; print(sklearn.__version__)"