def run_article_analysis():
# Connecting to the databse
engine = db_connect()
Session = sessionmaker(bind=engine)
session = Session()
for article in session.query(Articles):
articleAnalysis = DocumentAnalysis(article, False)
articleAnalysis.analysis()
results = {}
for issue in issues:
key = issue.get('key')
if key in summary['issues']:
content = summary['issues'][key].get('content')
examples = summary['issues'][key].get('examples')
readability = 0
sentiment = 0
for example in examples:
readability += example.get('readability')
sentiment += sentiment
content_words = get_lowercase(get_words(content))
word_count = len(content_words)
unique_word_count = len(set(content_words))
content_words_without_stopwords = remove_stop_words(content_words)
word_frequency = nltk.FreqDist(content_words_without_stopwords).most_common(20)
results[key] = {
'word_count': word_count,
'unique_word_count': unique_word_count,
'20_most_frequent': word_frequency,
'instances': len(examples),
'readability': readability/len(examples),
'sentiment': sentiment,
}
print(results)
def run_press_release_analysis():
# Connecting to the databse
engine = db_connect()
Session = sessionmaker(bind=engine)
session = Session()
for press_release in session.query(PressReleases):
press_release_analysis = DocumentAnalysis(press_release, True)
press_release_analysis.analysis()
try:
session.add(press_release_analysis.document)
session.commit()
except:
session.rollback()
def __init__(self):
engine = db_connect()
create_articles_table(engine)
self.Session = sessionmaker(bind=engine)
self.stats = {}