forked from newsWhisperer/winterIsComing
-
Notifications
You must be signed in to change notification settings - Fork 1
/
entities.py
222 lines (185 loc) · 10.2 KB
/
entities.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
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
import pandas as pd
from pathlib import Path
import os.path
import io
#import requests
import glob
import datetime
from dateutil import parser
# pip3 install spacy
# python3 -m spacy download de_core_news_md
#pip3 install textblob_de
import spacy
import de_core_news_md
from textblob_de import TextBlobDE
nlp = de_core_news_md.load()
DATA_PATH = Path.cwd()
if(not os.path.exists(DATA_PATH / 'csv')):
os.mkdir(DATA_PATH / 'csv')
def getNewsFiles():
fileName = './csv/news_????_??.csv'
files = glob.glob(fileName)
return files
def getNewsDFbyList(files):
newsDF = pd.DataFrame(None)
for file in files:
df = pd.read_csv(file, delimiter=',')
if(newsDF.empty):
newsDF = df
else:
newsDF = pd.concat([newsDF, df])
newsDF = newsDF.sort_values(by=['published'], ascending=True)
return newsDF
def getNewsDF():
files = getNewsFiles()
newsDF = getNewsDFbyList(files)
return newsDF
keywordsDF = pd.read_csv(DATA_PATH / 'keywords.csv', delimiter=',')
keywordsDF = keywordsDF.drop(columns = ['language'])
newsDf = getNewsDF()
print(newsDf)
keywordsNewsDF = pd.merge(keywordsDF, newsDf, how='left', left_on=['keyword'], right_on=['keyword'])
print(keywordsNewsDF)
newsDf['subjectivity'] = 0.0
newsDf['sentiment'] = 0.0
newsDf['count'] = 1.0
newsDf['week'] = '0000-00'
newsDf['day'] = '000-00-00'
i=0
##topicWordsAbs = {'summaryOfAllWords': emptyTopics.copy()}
for index, column in newsDf.iterrows():
i += 1
if(i % 50 == 0):
print(i)
quote = str(column.title)+'. ' +str(column.description)+' '+str(column.content)
#quote = str(column.title)+'. ' +str(column.description)
blob = TextBlobDE(quote)
newsDf.loc[newsDf['url'] == column['url'], 'subjectivity'] = blob.sentiment.subjectivity
newsDf.loc[newsDf['url'] == column['url'], 'sentiment'] = blob.sentiment.polarity
pubDate = parser.parse(column['published'])
newsDf.loc[newsDf['url'] == column['url'], 'week'] = pubDate.strftime('%Y-%W')
newsDf.loc[newsDf['url'] == column['url'], 'day'] = pubDate.strftime('%Y-%m-%d')
##keywordsNewsDF = newsDf.groupby('keyword').mean()
def groupSentiments(df, aggColumn):
cols = [aggColumn,'sentiment_mean','sentiment_std','subjectivity_mean','subjectivity_std','counting']
groupDF = df.groupby([aggColumn], as_index=False).agg(
{'sentiment':['mean','std'],'subjectivity':['mean','std'],'count':'sum'})
groupDF.columns = cols
groupDF.reindex(columns=sorted(groupDF.columns))
groupDF = groupDF.sort_values(by=['counting'], ascending=False)
groupDF['sentiment_std'] = groupDF['sentiment_std'].fillna(1)
groupDF['subjectivity_std'] = groupDF['subjectivity_std'].fillna(1)
return groupDF
domainDF = groupSentiments(newsDf, 'domain')
domainDF.loc[domainDF['counting'] < 2, 'sentiment_mean'] = 0.0
domainDF.loc[domainDF['counting'] < 2, 'subjectivity_mean'] = 0.0
print(domainDF)
cols = ['domain','sentiment_mean','sentiment_std','subjectivity_mean','subjectivity_std','counting']
domainDF.to_csv(DATA_PATH / 'csv' / 'sentiments_domains.csv', columns=cols,index=False)
objNewsDF = pd.merge(newsDf, domainDF, how='left', left_on=['domain'], right_on=['domain'])
objNewsDF['subjectivity'] = (objNewsDF['subjectivity'] - objNewsDF['subjectivity_mean'])/objNewsDF['subjectivity_std']
objNewsDF['sentiment'] = (objNewsDF['sentiment'] - objNewsDF['sentiment_mean'])/objNewsDF['sentiment_std']
print(objNewsDF)
weeksDF = groupSentiments(objNewsDF, 'week')
weeksDF = weeksDF.sort_values(by=['week'], ascending=True)
weeksDF.to_csv(DATA_PATH / 'csv' / 'sentiments_weeks.csv',index=False)
daysDF = groupSentiments(objNewsDF, 'day')
daysDF = daysDF.sort_values(by=['day'], ascending=True)
daysDF.to_csv(DATA_PATH / 'csv' / 'sentiments_days.csv',index=False)
keywordsSentimentDF = groupSentiments(objNewsDF, 'keyword')
keywordsSentimentDF = keywordsSentimentDF.sort_values(by=['keyword'], ascending=True)
keywordsSentimentDF.to_csv(DATA_PATH / 'csv' / 'sentiments_keywords.csv',index=False)
print(list(newsDf.columns))
print(list(objNewsDF.columns))
print(list(keywordsDF.columns))
topicNewsDF = pd.merge(objNewsDF, keywordsDF, how='left', left_on=['keyword'], right_on=['keyword'])
print(list(topicNewsDF.columns))
topicsDF = groupSentiments(topicNewsDF, 'topic')
topicsDF = topicsDF.sort_values(by=['topic'], ascending=True)
topicsDF.to_csv(DATA_PATH / 'csv' / 'sentiments_topics.csv',index=False)
emptyDict = {'count':0,'sentiment':0,'subjectivity':0}
indexLocations = {}
indexOrganizations = {}
indexPersons = {}
indexMisc = {}
indexMissing = {}
i=0
##topicWordsAbs = {'summaryOfAllWords': emptyTopics.copy()}
for index, column in objNewsDF.iterrows():
i += 1
if(i % 50 == 0):
print(i)
quote = str(column.title)+'. ' +str(column.description)+' '+str(column.content)
lang = column.language
#quote = str(column.title)+'. ' +str(column.description)
blob = TextBlobDE(quote)
for sentence in blob.sentences:
#sentence.sentiment.polarity
doc = nlp(str(sentence))
for entity in doc.ents:
if(entity.label_ in ['LOC','GPE']):
if(entity.text in indexLocations):
indexLocations[entity.text]['count'] += 1
indexLocations[entity.text]['sentiment'] += sentence.sentiment.polarity
indexLocations[entity.text]['subjectivity'] += sentence.sentiment.subjectivity
else:
indexLocations[entity.text] = {'phrase':entity.text, 'label':entity.label_, 'sentiment':sentence.sentiment.polarity,
'subjectivity':sentence.sentiment.subjectivity, 'language':lang,'count':1}
elif(entity.label_ in ['PER','PERSON']):
if(entity.text in indexPersons):
indexPersons[entity.text]['count'] += 1
indexPersons[entity.text]['sentiment'] += sentence.sentiment.polarity
indexPersons[entity.text]['subjectivity'] += sentence.sentiment.subjectivity
else:
indexPersons[entity.text] = {'phrase':entity.text, 'label':entity.label_, 'sentiment':sentence.sentiment.polarity,
'subjectivity':sentence.sentiment.subjectivity, 'language':lang, 'count':1}
elif('ORG' == entity.label_):
if(entity.text in indexOrganizations):
indexOrganizations[entity.text]['count'] += 1
indexOrganizations[entity.text]['sentiment'] += sentence.sentiment.polarity
indexOrganizations[entity.text]['subjectivity'] += sentence.sentiment.subjectivity
else:
indexOrganizations[entity.text] = {'phrase':entity.text, 'label':entity.label_, 'sentiment':sentence.sentiment.polarity,
'subjectivity':0, 'language':lang, 'count':1}
elif('MISC' == entity.label_):
if(entity.text in indexMisc):
indexMisc[entity.text]['count'] += 1
indexMisc[entity.text]['sentiment'] += sentence.sentiment.polarity
indexMisc[entity.text]['subjectivity'] += sentence.sentiment.subjectivity
else:
indexMisc[entity.text] = {'phrase':entity.text, 'label':entity.label_, 'sentiment':sentence.sentiment.polarity,
'subjectivity':sentence.sentiment.subjectivity, 'language':lang, 'count':1}
else:
if(entity.text in indexMissing):
indexMissing[entity.text]['count'] += 1
indexMissing[entity.text]['sentiment'] += sentence.sentiment.polarity
indexMissing[entity.text]['subjectivity'] += sentence.sentiment.subjectivity
else:
indexMissing[entity.text] = {'phrase':entity.text, 'label':entity.label_, 'sentiment':sentence.sentiment.polarity,
'subjectivity':sentence.sentiment.subjectivity, 'language':lang, 'count':1}
colSent = ['phrase', 'label', 'sentiment', 'subjectivity', 'language', 'count']
indexLocationsDF = pd.DataFrame.from_dict(indexLocations, orient='index', columns=colSent)
indexLocationsDF['sentiment'] = indexLocationsDF['sentiment']/indexLocationsDF['count']
indexLocationsDF['subjectivity'] = indexLocationsDF['subjectivity']/indexLocationsDF['count']
indexLocationsDF = indexLocationsDF.sort_values(by=['count'], ascending=False)
indexLocationsDF.to_csv(DATA_PATH / 'csv' / "sentiments_locations.csv", index=True)
indexPersonsDF = pd.DataFrame.from_dict(indexPersons, orient='index', columns=colSent)
indexPersonsDF['sentiment'] = indexPersonsDF['sentiment']/indexPersonsDF['count']
indexPersonsDF['subjectivity'] = indexPersonsDF['subjectivity']/indexPersonsDF['count']
indexPersonsDF = indexPersonsDF.sort_values(by=['count'], ascending=False)
indexPersonsDF.to_csv(DATA_PATH / 'csv' / "sentiments_persons.csv", index=True)
indexOrganizationsDF = pd.DataFrame.from_dict(indexOrganizations, orient='index', columns=colSent)
indexOrganizationsDF['sentiment'] = indexOrganizationsDF['sentiment']/indexOrganizationsDF['count']
indexOrganizationsDF['subjectivity'] = indexOrganizationsDF['subjectivity']/indexOrganizationsDF['count']
indexOrganizationsDF = indexOrganizationsDF.sort_values(by=['count'], ascending=False)
indexOrganizationsDF.to_csv(DATA_PATH / 'csv' / "sentiments_organizations.csv", index=True)
indexMiscDF = pd.DataFrame.from_dict(indexMisc, orient='index', columns=colSent)
indexMiscDF['sentiment'] = indexMiscDF['sentiment']/indexLocationsDF['count']
indexMiscDF['subjectivity'] = indexMiscDF['subjectivity']/indexLocationsDF['count']
indexMiscDF = indexMiscDF.sort_values(by=['count'], ascending=False)
indexMiscDF.to_csv(DATA_PATH / 'csv' / "sentiments_misc.csv", index=True)
indexMissingDF = pd.DataFrame.from_dict(indexMissing, orient='index', columns=colSent)
indexMissingDF['sentiment'] = indexMissingDF['sentiment']/indexLocationsDF['count']
indexMissingDF['subjectivity'] = indexMissingDF['subjectivity']/indexLocationsDF['count']
indexMissingDF = indexMissingDF.sort_values(by=['count'], ascending=False)
indexMissingDF.to_csv(DATA_PATH / 'csv' / "sentiments_missing.csv", index=True)