-
Notifications
You must be signed in to change notification settings - Fork 10
/
lda.py
189 lines (156 loc) · 8.22 KB
/
lda.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
# -*- coding:utf-8 -*-
"""
Create on 2020/9/4 11:08 上午
@Author: dfsj
@Description: LDA 主题模型
"""
import pandas as pd
import gensim
from gensim.corpora import Dictionary
from gensim.models import CoherenceModel
import matplotlib.pyplot as plt
from gensim.models.wrappers import LdaMallet
from config import *
pd.set_option('display.max_columns', None)
pd.set_option('display.max_rows', None)
pd.set_option('max_colwidth', 80)
def draw_train_result(x, coherence_values):
# Print the coherence scores
for m, cv in zip(x, coherence_values):
logger.info("Num Topics =", m, " has Coherence Value of", round(cv, 4))
plt.plot(x, coherence_values)
plt.xlabel("Num Topics")
plt.ylabel("Coherence score")
plt.legend("coherence_values", loc='best')
plt.savefig(PIC_LDA)
def get_best_model_index(x, coherence_values: list):
"""
选择最佳模型，目前下方自动化选择最佳模型的想法还不够成熟，此处目前最好人工选择模型
实际上最好的不一定是 coherence_value 得分最高的，而是较为平滑曲线的左侧端点，
此处为了简单，选择具有最高 coherence_value 的模型
"""
for m, cv in zip(x, coherence_values):
logger.info("Num Topics =", m, " has Coherence Value of", round(cv, 4))
max_value = max(coherence_values)
max_value_index = coherence_values.index(max_value)
# index = max_value_index
# while index >= 0 and abs(coherence_values[index - 1] - coherence_values[index]) < 0.002:
# index -= 1
# return index
return max_value_index
class LDA:
best_model: LdaMallet
id2word: Dictionary
corpus: list
def __init__(self, texts, num_topics=None, start=5, limit=15, step=5, print_summary=True, workers=1):
"""
LDA 模型
:param texts: 经过分词后的文本表示形式
:param num_topics: 选定主题数，如果提供该值，则不进行超参搜索
:param start: 当不提供主题数，根据指定的超参范围进行搜索，起始
:param limit: 终止
:param step: 步长
"""
self.num_topics = num_topics
self.start = start
self.limit = limit
self.step = step
self.texts = texts
self.print_summary = print_summary
self.workers = workers
self.data_process()
def data_process(self):
logger.info("数据转化中，请稍后 ...")
self.id2word = Dictionary(self.texts) # 生成字典
self.id2word.save_as_text(DICT_NAME) # 保存字典
logger.info("字典生成完毕，示例：id2word[0] = {}".format(self.id2word[0]))
self.corpus = [self.id2word.doc2bow(text) for text in self.texts] # 生成 LDA 所需语料形式
logger.info("数据转化完成，示例：corpus[0] = {}".format(self.corpus[0]))
logger.info("人为可读转化，corpus[0] = {}".format([(self.id2word[index], freq) for index, freq in self.corpus[0]]))
def compute_coherence_values(self, limit, start=2, step=3):
"""
Compute c_v coherence for various number of topics
Parameters:
----------
dictionary : Gensim dictionary
corpus : Gensim corpus
texts : List of input texts
limit : Max num of topics
Returns:
-------
model_list : List of LDA topic models
coherence_values : Coherence values corresponding to the LDA model with respective number of topics
"""
coherence_values = []
model_list = []
for num_topics in range(start, limit, step):
model = gensim.models.wrappers.LdaMallet(
PATH_MALLET, corpus=self.corpus, num_topics=num_topics, id2word=self.id2word, workers=self.workers)
model_list.append(model)
coherencemodel = CoherenceModel(model=model, texts=self.texts, dictionary=self.id2word, coherence='c_v')
coherence_values.append(coherencemodel.get_coherence())
return model_list, coherence_values
def train(self):
if self.num_topics:
self.best_model = gensim.models.wrappers.LdaMallet(
PATH_MALLET, corpus=self.corpus, num_topics=self.num_topics,
id2word=self.id2word, workers=self.workers, prefix=PREFIX)
coherencemodel = CoherenceModel(
model=self.best_model, texts=self.texts, dictionary=self.id2word, coherence='c_v')
coherence_ldamallet = coherencemodel.get_coherence()
logger.info('Coherence Score: {}'.format(coherence_ldamallet))
else:
model_list, coherence_values = self.compute_coherence_values(start=5, limit=15, step=5)
x = range(self.start, self.limit, self.step)
draw_train_result(x, coherence_values) # 画图
# ****************************************************************************************
self.best_model = model_list[get_best_model_index(x, coherence_values)]
# 目前自动化选择最佳模型的想法还不成熟
# ****************************************************************************************
self.print_topic()
if self.print_summary:
self.get_summary()
logger.info(self.best_model[self.corpus[0]])
return self.best_model
def print_topic(self):
# model_topics = self.best_model.show_topics(formatted=False)
logger.info(self.best_model.print_topics(num_words=10))
def format_topics_sentences(self):
""" 对每一条数据进行主题标注，并给出该主题的百分比以及关键词内容等 """
sent_topics_df = pd.DataFrame()
# Get main topic in each document
for i, row in enumerate(self.best_model[self.corpus]):
row = sorted(row, key=lambda x: (x[1]), reverse=True)
# Get the Dominant topic, Perc Contribution and Keywords for each document
for j, (topic_num, prop_topic) in enumerate(row):
if j == 0: # => dominant topic
wp = self.best_model.show_topic(topic_num)
topic_keywords = ", ".join([word for word, prop in wp])
sent_topics_df = sent_topics_df.append(
pd.Series([int(topic_num), round(prop_topic, 4), topic_keywords]), ignore_index=True)
else:
break
sent_topics_df.columns = ['Dominant_Topic', 'Perc_Contribution', 'Topic_Keywords']
contents = pd.Series(self.texts)
sent_topics_df = pd.concat([sent_topics_df, contents], axis=1)
return sent_topics_df
def get_summary(self):
""" 打印统计信息 """
sent_topics_sorteddf_mallet = pd.DataFrame()
df_topic_sents_keywords = self.format_topics_sentences() # 得到每一条数据的主题信息
sent_topics_outdf_grpd = df_topic_sents_keywords.groupby('Dominant_Topic') # 以主题分组
for _, grp in sent_topics_outdf_grpd:
sent_topics_sorteddf_mallet = pd.concat(
[sent_topics_sorteddf_mallet,
grp.sort_values(['Perc_Contribution'], ascending=[0]).head(1)], axis=0) # 只选择贡献度最高的一个
sent_topics_sorteddf_mallet.reset_index(drop=True, inplace=True)
sent_topics_sorteddf_mallet.columns = ['Topic_Num', "Topic_Perc_Contrib", "Keywords", "Text"]
topic_counts = df_topic_sents_keywords['Dominant_Topic'].value_counts() # 每个主题的文档数
topic_contribution = round(topic_counts / topic_counts.sum(), 4) # 每个主题的文档占比
topic_num_keywords = df_topic_sents_keywords[
['Dominant_Topic', 'Topic_Keywords']].drop_duplicates().reset_index(drop=True)
df_dominant_topics = pd.concat([topic_num_keywords, topic_counts, topic_contribution], axis=1)
df_dominant_topics.columns = ['Topic_Num', 'Keywords', 'Num_Documents', 'Perc_Documents']
df_dominant_topics.sort_values(by="Topic_Num", ascending=True, inplace=True)
summary = pd.merge(sent_topics_sorteddf_mallet, df_dominant_topics, on=["Topic_Num", "Keywords"])
logger.info(summary)