题目：二元语法模型
进行参数的初始化操作，如无重复字典，和字典索引，所有的句子列表等
#统计每个单词的出现次数用字典标注,参数返回一个字典
def count_word_hapeen(text):
for sentence in text:
notDeal_sentence = sentence.split('|')[2].strip('\n').translate(str.maketrans('', '', string.punctuation))
word_sentence_list.append(notDeal_sentence)
for word in notDeal_sentence.split(" "):
word_list.append(word)
#开始向字典里面增加数据，并开始计数
count = 0
for i in word_list:
if(word_dir.get(i)!=None):
word_dir[i]=word_dir[i]+1
# word_dir.update(i,word_dir.values(i)+1)
else:
word_dir_sl[i]=count
count+=1
# print(word_dir.items())
print('\n')
word_dir[i]=1
#通过这里删选的时候来进而存放无重复列表单词
word_unRepeat_list.append(i)
进行二维表的创建，上面的属于高效方法，下面那个太慢
#开始对文本里面的字符进行组合判断出现次数
def make_word(word_array):
print("\n")
print('length',len(word_dir))
count_number=0
#还是得看这里
for k in word_sentence_list:
sen_had_split = k.split()
for h in range(0, (len(k.split()) - 1)):
# print('目前下标',sen_had_split[h],' '+sen_had_split[h-1],'\n')
# print('字典里面的坐标',word_dir_sl[sen_had_split[h]], ' ' ,word_dir_sl[sen_had_split[h-1]], '\n')
word_1 = word_dir_sl[sen_had_split[h]];
word_2 = word_dir_sl[sen_had_split[h-1]];
word_array[word_1][word_2]+=1
#print('1下标',word_1,'2下标',word_2,'\n')
return word_array
'''
下面的蠢方法,还是别弄了
for i in range(0,len(word_dir)):
for j in range(0,len(word_dir)):
#开始对i j 来进行判断
for k in word_sentence_list:
#开始针对第二次方式，
sen_had_split = k.split()
for h in range(0,(len(k.split())-1)):
#逐行进行判断
if((word_unRepeat_list[i]==sen_had_split[h])and( word_unRepeat_list[j]==sen_had_split[h+1])):
count_number = count_number+1
word_array[i][j] = count_number
if(count_number>0):
print('\n')
print('i行=',i,'j列=',j, word_array[i][j])
count_number = 0
'''
二维表的概率表建立公式
word_problity_array[i][j] =float((word_array[i][j])/ (word_dir[word_unRepeat_list[i]]))
函数如下：
#开始计算每个出现的概率公式P（i，j）= word_array[i][j]/list[i]
def calculation_Array_Probability(word_array):
word_problity_array = np.zeros((len(word_dir), len(word_dir)), dtype=float) # 存储二维数据表
for i in range(0,len(word_array)):
print('\n')
for j in range(0,len(word_array)):
word_problity_array[i][j] =float((word_array[i][j])/ (word_dir[word_unRepeat_list[i]]))
print('data ', word_problity_array[i][j],' ',word_dir[word_unRepeat_list[i]])
return word_problity_array
第二部分问题
1、函数输入问题
def cycle_input_word():
flag = True
while(flag):
input_word = ''
choice_number = -1
wait_slect_list = []
user_input_word = input()
if(user_input_word==-1):
flag = False
else:
if(word_dir.get(input_word)==None):
wait_slect_list = []
else:
wait_slect_list = find_five_word_list(input())
print("choice 1、2、3、4、5")
print("\n")
for i in wait_slect_list:
print(i," ")
choice_number = input()
user_input_word = wait_slect_list[choice_number]
2、找出一列最高的五个数字,返回一个list
def find_five_word_list(word):
five_list = []
index = word_list.index(word)
word_find_one_array = np.zeros((len(word_dir)), dtype=int)
for i in range(0,len(word_array)):
word_find_one_array[i] = word_array[index][i]
#赋值完毕，建立堆来直接找出最多的五个单词
index_list = map(word_find_one_array.index, heapq.nlargest(5, word_find_one_array))
for j in range(0,len(index)):
five_list.append(word_array[index][index_list[j]])
return five_list
完整如下
import numpy as np
import string
import heapq
import random
word_list = []#存放所有单词
word_dir = {}#存放无重复字典
word_dir_sl = {}#存放字典索引
word_sentence_list = []#存放句子一维表
word_unRepeat_list = []#存放无重复列表单词
#统计每个单词的出现次数用字典标注,参数返回一个字典
def count_word_hapeen(text):
for sentence in text:
notDeal_sentence = sentence.split('|')[2].strip('\n').translate(str.maketrans('', '', string.punctuation))
word_sentence_list.append(notDeal_sentence)
for word in notDeal_sentence.split(" "):
word_list.append(word)
#开始向字典里面增加数据，并开始计数
count = 0
for i in word_list:
if(word_dir.get(i)!=None):
word_dir[i]=word_dir[i]+1
# word_dir.update(i,word_dir.values(i)+1)
else:
word_dir_sl[i]=count
count+=1
# print(word_dir.items())
# print('\n')
word_dir[i]=1
#通过这里删选的时候来进而存放无重复列表单词
word_unRepeat_list.append(i)
# print(word_dir.items())
#开始对文本里面的字符进行组合判断出现次数
def make_word(word_array):
#print("\n")
#print('length',len(word_dir))
count_number=0
#还是得看这里
for k in word_sentence_list:
sen_had_split = k.split()
for h in range(0, (len(k.split()) - 2)):
# print('目前下标',sen_had_split[h],' '+sen_had_split[h-1],'\n')
# print('字典里面的坐标',word_dir_sl[sen_had_split[h]], ' ' ,word_dir_sl[sen_had_split[h-1]], '\n')
word_1 = word_dir_sl[sen_had_split[h]];
word_2 = word_dir_sl[sen_had_split[h+1]];
word_array[word_1][word_2]+=1
return word_array
'''
下面的蠢方法,还是别弄了
for i in range(0,len(word_dir)):
for j in range(0,len(word_dir)):
#开始对i j 来进行判断
for k in word_sentence_list:
#开始针对第二次方式，
sen_had_split = k.split()
for h in range(0,(len(k.split())-1)):
#逐行进行判断
if((word_unRepeat_list[i]==sen_had_split[h])and( word_unRepeat_list[j]==sen_had_split[h+1])):
count_number = count_number+1
word_array[i][j] = count_number
if(count_number>0):
print('\n')
print('i行=',i,'j列=',j, word_array[i][j])
count_number = 0
'''
def printData(word_array):
print("length",len(word_dir))
for i in range(0,len(word_dir)-1):
print("\n")
for j in range(0,len(word_dir)-1):
print('data ',word_array[i][j],end=' ')
#开始计算每个出现的概率公式P（i，j）= word_array[i][j]/list[i]
def calculation_Array_Probability(word_array):
word_problity_array = np.zeros((len(word_dir), len(word_dir)), dtype=float) # 存储二维数据表
for i in range(0,len(word_array)):
print('\n')
for j in range(0,len(word_array)):
word_problity_array[i][j] =float((word_array[i][j])/ (word_dir[word_unRepeat_list[i]]))
print('data ', word_problity_array[i][j],' ',word_dir[word_unRepeat_list[i]])
return word_problity_array
'''
第二部分问题函数
'''
def ge_position_index(x):
return
def find_five_word_list(word,array):
five_list = []
index = word_dir_sl[word]
test = [0]*len(word_array)
# word_find_one_array = np.zeros(len(word_dir), dtype=int)
for i in range(0,len(word_array)):
test[i] = array[index][i]
#赋值完毕，建立堆来直接找出最多的五个单词
index_list = map(test.index,heapq.nlargest(5, test))
temper_dir = {}
for el in index_list:
if (temper_dir.get(el) != None):
five_list.append(word_unRepeat_list[random.randint(0,len(word_unRepeat_list))])
else:
temper_dir[el] = 1
five_list.append(word_unRepeat_list[el])
'''
for j in range(0,5):
five_list.append(index_list[j])
'''
return five_list
def cycle_input_word(word_array):
flag = True
user_input_word = input()
while(flag):
input_word = ''
choice_number = -1
wait_slect_list = []
if(user_input_word==-1):
flag = False
else:
if(word_dir.get(input_word)==None):
wait_slect_list = []
else:
wait_slect_list = find_five_word_list(user_input_word,word_array)
print("点击数字 1、2、3、4、5")
for i in wait_slect_list:
print(" ",i," ")
choice_number = input()
user_input_word = wait_slect_list[int(choice_number)]
if __name__ == '__main__':
#开始一些必要的函数，对全局变量的初始化
#word_array=make_word();
text = open('metadata.txt',mode='r',encoding='UTF-8');
# 初始化变量word_list和word_sentence_list
#splitText(text)
count_word_hapeen(text)
word_array = np.zeros((len(word_dir), len(word_dir)), dtype=float) # 存储二维数据表
#制作二维表实例
word_problity_array = np.zeros((len(word_dir),len(word_dir)),dtype=int)#存储二维数据表
word_array = make_word(word_array)
#printData(word_array)
#开始对二维表进行计算其MLE，先计算每个单独的概率P（i，j）= word_array[i][j]/list[i]
#word_problity_array = calculation_Array_Probability(word_array)
'''
开始第二部分问题，由输入数据input()来进行补充，接着开始推荐几个后面跟着的候补单词
直接从，该单词行来找出，前五个最大的数->从word_array取直接放入list。输出
由于循环输出，直接使用while(),输入截至的标志为-1；
'''
print("---------------请输入单词----------")
cycle_input_word(word_array)
printData(word_array)
结语：共勉
225
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
