信息内容安全-朴素贝叶斯邮件过滤（多项式模型）实验

信息内容安全-朴素贝叶斯邮件过滤（多项式模型）实验

代码

# 信息内容安全作业-朴素贝叶斯邮件过滤（多项式模型）

import os
import jieba
import random

from collections import Counter
import numpy as np
from sklearn.naive_bayes import MultinomialNB, GaussianNB, BernoulliNB
from sklearn.svm import SVC, NuSVC, LinearSVC
from sklearn.metrics import confusion_matrix


# 提取 50 封邮件并标注好类型（正常邮件为1）
def get_email():
    global email_list
    global email_num

    # 提取正常邮件，并标注类型为1
    for root, dirs, files in os.walk(ham_path):
        for fn in files:
            email_list.append([fn,1])
            email_num += 1

    # 提取垃圾邮件，并标注类型为0
    for root, dirs, files in os.walk(spam_path):
        for fn in files:
            email_list.append([fn,0])
            email_num += 1


# 从 spam 和 pam 集中随机取 10 封邮件作为测试集，剩下 40 封作为训练集
# （并标注类型、记录各类型邮件数量）
def get_training_set(email_list):
    global training_set
    global test_set
    global ham_training_num
    global spam_training_num
    global email_training_num
    training_set = email_list[:]
    email_training_num = email_num

    for i in range(10):
        s = int( random.uniform(0, len(training_set)) )
        if training_set[s][1] == 1:
            ham_training_num -= 1
        else:
            spam_training_num -= 1
        email_training_num -= 1
        test_set.append(training_set[s])
        del (training_set[s])

    print("测试集：")
    print(test_set)


# 计算先验概率 P(y)
def get_priori_possibility():
    global py1
    global py2

    py1 = round( (ham_training_num + a) / (email_training_num + 2 * a), 2)
    py2 = round( (spam_training_num + a) / (email_training_num +