-
Notifications
You must be signed in to change notification settings - Fork 1
/
pipeline.py
167 lines (146 loc) · 7.49 KB
/
pipeline.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
#!/usr/bin/env python3.8
"""
Central project pipeline, controlled by a config.ini file.
Calls functions for preprocessing and analysis.
"""
import configparser
import glob
import sys
import article_selection.article_selection as article_selection
from sentiment_analysis.inference import calulate_sentiment, eval_sentiment
from sentiment_analysis.word2vec_sentiment import *
from visualization.dash_plot import dash_plot
from visualization.wordcloud import generate_word_clouds
if __name__ == "__main__":
# read config file
try:
config_file = sys.argv[1]
config = configparser.ConfigParser()
config.read(config_file)
except IndexError:
print("Error: Supply a config.ini file")
quit()
# ==============================
# Selection of relevant Articles
# ==============================
if config.getboolean("ArticleSelection", "run_article_selection"):
# create input filepath for article selection from:
# the path to the folders and the start and end year
base_path = config.get("ArticleSelection", "input_path_base")
start_year = config.getint("ArticleSelection", "start_year")
end_year = config.getint("ArticleSelection", "end_year")
data_path_list = [base_path + str(year) + "/" for year in range(start_year, end_year + 1)]
# create list of all data paths
json_file_list = []
for path in data_path_list:
json_file_list += [file_path for file_path in glob.glob(path + "*.json")]
# get keywords, output_path
search_keywords = config.get("ArticleSelection", "search_words").lower().split(", ")
output_base = config.get("ArticleSelection", "output_base")
# create new file or append to existing file
create_new_files = not config.getboolean("ArticleSelection", "append_to_existing_file")
# if annotation files are needed get training_size and seed
use_annotation = config.getboolean("ArticleSelection", "use_annotation")
training_size = config.getint("ArticleSelection", "training_size")
seed = config.getint("ArticleSelection", "seed")
# open all files containing an article
# check if the topic is relevannt
# if use_anotation is True output is split in four files:
# 1. evaluation (size_all-training_size)
# 2. 3 annotation files with the names of the annotators 1/3 training_size
article_selection.write_relevant_content_to_file(json_file_list,
output_base,
search_keywords=search_keywords,
new=create_new_files,
training_size=training_size,
seed=seed,
annotation=use_annotation)
# ===================
# Word2Vec analysis
# ===================
if config.getboolean("Analysis", "run_w2v"):
print("\nStart word2vec analysis")
input_file = config.get("Analysis", "input_file")
search_words = config.get("Analysis", "search_words_w2v").lower().split(",")
base_output_path = config.get("Analysis", "output_base_w2v")
start_year = config.getint("Analysis", "start_year")
end_year = config.getint("Analysis", "end_year")
number_most_sim = config.getint("Analysis", "number_most_sim")
# all articles of the same year are one dataset
if config.getboolean("Analysis", "run_by_year"):
similarity_by_year(input_file, base_output_path, search_words,
start_year, end_year, number_most_sim)
# all articles of the same publisher are one dataset
if config.getboolean("Analysis", "run_by_publisher"):
similarity_by_publisher(input_file, base_output_path, search_words,
start_year, end_year, number_most_sim)
# all articles of the same publisher during the same year are one dataset
if config.getboolean("Analysis", "run_by_publisher_by_year"):
similarity_by_year_and_publisher(input_file, base_output_path,
search_words, start_year,
end_year, number_most_sim)
print("\nEnd word2vec analysis\n")
# ==================
# Sentiment analysis
# ==================
if config.getboolean("Analysis", "run_senti"):
print("\nStart sentiment analysis")
input_file = config.get("Analysis", "input_file")
search_words = config.get("Analysis", "search_words").lower().split(",")
output_file = config.get("Analysis", "output_senti")
methods = config.get('Analysis', 'senti_methods').lower().split(", ")
finetuned_sentibert_path = config.get('Analysis', 'finetuned_sentibert_path')
# calculate the article sentiment
# using one or multiple of the following methods:
# 1. Sentiment-Dictionary (sentiws)
# 2. Neuronal-Network based Bert model
# trained for gerneral sentiment analysis
# 3. Same Bert model with additional training
# using labeled parts of news-articles about refugees
calulate_sentiment(
input_file,
output_file,
search_words,
methods=methods,
finetuned_sentibert_path=finetuned_sentibert_path
)
# =============================
# Sentiment analysis evaluation
# =============================
if config.getboolean("Analysis", "run_senti_eval"):
print("\nStart Evaluation")
senti_eval_input = config.get("Analysis", "senti_eval_input")
search_words = config.get("Analysis", "search_words").lower().split(",")
senti_eval_output = config.get("Analysis", "senti_eval_output")
methods = config.get('Analysis', 'senti_methods').lower().split(", ")
finetuned_sentibert_path = config.get('Analysis', 'finetuned_sentibert_path')
# Perform quantitative evaluation of sentiment analysis approaches
# using one or multiple of the following methods:
# 1. Sentiment-Dictionary (sentiws)
# 2. Neuronal-Network based Bert model
# trained for gerneral sentiment analysis
# 3. Same Bert model with additional training
# using labeled parts of news-articles about refugees
eval_sentiment(
senti_eval_input,
senti_eval_output,
search_words,
methods=methods,
finetuned_sentibert_path=finetuned_sentibert_path
)
# ==================
# Plotting
# ==================
if config.getboolean("Plotting", "sentiment_plot"):
input_file = config.get("Plotting", "input_file")
dash_plot(input_file)
# ==================
# WordClouds
# ==================
if config.getboolean("WordClouds", "wordcloud_plot"):
input_file = config.get("WordClouds", "input_file")
output_path = config.get("WordClouds", "output_path")
words = config.get("WordClouds", "words").lower().split(", ")
column_values = config.get("WordClouds", "column_values").lower().split(", ")
number_of_words_in_wordcloud = config.getint("WordClouds", "number_of_words_in_wordcloud")
generate_word_clouds(input_file, words, column_values, output_path, number_of_words_in_wordcloud)