HMM模型下的中文分词的Python3代码

参数估计代码：

# -*- coding: utf-8 -*-

# 二元隐马尔科夫模型（Bigram HMMs）

# 'trainText.txt_utf8'为人民日报已经人工分词的预料，29万多条句子

下载地址：

链接：https://pan.baidu.com/s/1kXosD1P 密码：mq61

import sys


def getList(input_str):
    outpout_str = []
    if len(input_str) == 1:
        outpout_str.append('S')
    elif len(input_str) == 2:
        outpout_str = ['B', 'E']
    else:
        M_num = len(input_str) - 2
        M_list = ['M'] * M_num
        outpout_str.append('B')
        outpout_str.extend(M_list)
        outpout_str.append('E')
    return outpout_str




#state_M = 4
#word_N = 0
A_dic = {}
B_dic = {}
Count_dic = {}   #number of all emerged state
Pi_dic = {}      #number of begin state
word_set = set()


state_list = ['B', 'M', 'E', 'S']


line_num = -1


INPUT_DATA = "trainText.txt_utf8"
PROB_START = "trainHMM\prob_start.py"
PROB_EMIT = "trainHMM\prob_emit.py"
PROB_TRANS = "trainHMM\prob_trans.