def send_to_indico():
'''
This route handles the server's response when
you post data to localhost:5000/crunch through
the form on index.html
'''
tweets_csv_string = request.form.get('tweets')
csv_list = tweets_csv_string.replace('\r', '').splitlines()
if len(csv_list) > 40:
csv_list = csv_list[:40]
tweet_list = []
for csv_tweet in csv_list:
tweet_only = csv_tweet.split(',')[2:]
tweet_list.append(','.join(tweet_only))
tweet_list = tweet_list[::-1]
tweet_scores = indicoio.batch_sentiment(tweet_list, api_key="YOUR_API_KEY")
return json.dumps({
'scores': tweet_scores,
'tweets': tweet_list
}) # dumps converts res to a JSON object
def get_indico_features(posts):
texts = [ post.title.encode('UTF8') for post in posts]
print "Getting sentiment scores..."
sentiment = indicoio.batch_sentiment(texts)
print "Getting political scores..."
political = indicoio.batch_political(texts)
return (sentiment, political)
def send_to_indico():
'''
This route handles the server's response when
you post data to localhost:5000/crunch through
the form on index.html
'''
tweets_csv_string = request.form.get('tweets')
csv_list = tweets_csv_string.replace('\r', '').splitlines()
if len(csv_list) > 40:
csv_list = csv_list[:40]
tweet_list = []
for csv_tweet in csv_list:
tweet_only = csv_tweet.split(',')[2:]
tweet_list.append(','.join(tweet_only))
tweet_list = tweet_list[::-1]
tweet_scores = indicoio.batch_sentiment(tweet_list, api_key="YOUR_API_KEY")
return json.dumps({'scores': tweet_scores, 'tweets': tweet_list}) # dumps converts res to a JSON object
def post(self):
query_string = self.request.body_arguments.get('query')
query = TwitterSearchOrder()
query.set_keywords(query_string)
query.set_language('en')
query.set_include_entities(False)
results = TwitterClient.search_tweets(query)
tweets = [tweet['text'] for tweet in results['content']['statuses']]
sentiment = indicoio.batch_sentiment(tweets)
pairs = sorted(zip(sentiment, tweets))
n_tweets = float(len(pairs))
top_n = 5
most_negative = pairs[:top_n]
most_positive = list(reversed(pairs[-top_n:]))
data = {
'most_positive': most_positive,
'most_negative': most_negative,
'average': sum(sentiment)/n_tweets
}
self.write(json.dumps(data))
def post(self):
query_string = self.request.body_arguments.get('query')
query = TwitterSearchOrder()
query.set_keywords(query_string)
query.set_language('en')
query.set_include_entities(False)
results = TwitterClient.search_tweets(query)
tweets = [tweet['text'] for tweet in results['content']['statuses']]
sentiment = indicoio.batch_sentiment(tweets)
pairs = sorted(zip(sentiment, tweets))
n_tweets = float(len(pairs))
top_n = 5
most_negative = pairs[:top_n]
most_positive = list(reversed(pairs[-top_n:]))
data = {
'most_positive': most_positive,
'most_negative': most_negative,
'average': sum(sentiment) / n_tweets
}
self.write(json.dumps(data))
import indicoio
import pandas as pd
import csv
step_size = 200
indicoio.config.api_key = 'cefae61c6968c29ea880d918e9a6095d'
key2 = 'bff000095c5eb591fa1b4ab3da98063b'
df = pd.read_csv('data/hotel_reviews.csv')
reviews = df['text'].tolist()
sentiment = []
steps = len(sentiment) / step_size
leftover = len(sentiment) % step_size
total_count = 0
for i in xrange(steps):
total_count += step_size
if total_count > 20000:
indicoio.config.api_key = 'bff000095c5eb591fa1b4ab3da98063b'
test = reviews[i * step_size:(i + 1) * step_size]
sentiment.extend(indicoio.batch_sentiment(test))
test = reviews[total_count:]
sentiment.extend(indicoio.batch_sentiment(test))
csv_sent = [[x] for x in sentiment]
resultFile = open("sentiment_results.csv",'wb')
wr = csv.writer(resultFile, dialect='excel')
wr.writerows(csv_sent)
from indicoio import config, sentiment, batch_sentiment
import csv
config.api_key = "0285e71c8c80f055c102ad5f3a614fc2"
tweets=[]
with open('twitter_data.csv', 'rb') as f:
reader = csv.reader(f)
for row in reader:
tweets.append(row[1])
del(tweets[0])
batch = batch_sentiment(tweets)
for element in batch:
if(element>0.5):
print "Postive",
print ",",
else:
print "Negative",
print ",",
print element
#!/usr/bin/env python
# actually run the sentiment
import indicoio
import os
import glob
indicoio.config.api_key = 'oops'
import glob
for player in ['deandre','lamarcus']:
opinions_in_threads = {}
for files in glob.glob("/home/akshay/indicoio/{}/*.txt".format(player)):
with open(files,'r') as f:
comments = [line.decode('utf8') for line in f]
sentiments = indicoio.batch_sentiment(comments)
pos_sentiments = [sentiment for sentiment in sentiments if sentiment >=0.5]
neg_sentiments = [sentiment for sentiment in sentiments if sentiment <0.5]
pos_avg = sum(pos_sentiments)/(len(pos_sentiments) + 0.0)
neg_avg = sum(neg_sentiments)/(len(neg_sentiments) + 0.0)
opinions_in_threads[files.split('/')[-1].split('.')[0]] = (len(pos_sentiments),len(neg_sentiments),pos_avg,neg_avg)
with open('results.txt.{}'.format(player),'w') as f:
for sentiment in opinions_in_threads:
f.write('Thread: {}, '.format(sentiment) + 'Pos Comments: %d, Neg Comments: %d, Avg Pos: %f, Avg Neg: %f\n' % opinions_in_threads[sentiment])
def test_batch_posneg(self):
test_data = ['Worst song ever', 'Best song ever']
response = batch_sentiment(test_data, api_key=self.api_key)
self.assertTrue(isinstance(response, list))
self.assertTrue(response[0] < 0.5)
import indicoio
import numpy as np
import matplotlib.pyplot as plt
# This is an example of how the batch API works on three different words
indicoio.batch_sentiment(["Happy", "Terrible", "Wonderful"])
text_file = open("call_of_the_wild.txt", "r")
chunks = text_file.read().split()
text_file.close()
# We then slice the chunks to send to indico's sentiment API:
def chunker(seq, size):
return (seq[pos:pos + size] for pos in xrange(0, len(seq), 1))
# Split the data up into 30,000+ chunks of 100 words each
data = [" ".join(group) for group in chunker(chunks, 100)]
# Sending off data for remote processing
sentiment_scores = indicoio.batch_sentiment(data)
# Conversion to numpy arrays
x = np.arange(len(data))
y = np.array(sentiment_scores)
# Plotting and formatting
plt.scatter(x, y)
plt.ylim(ymax=1.0, ymin=0.0)
plt.xlim(xmax=325, xmin=-1.0)
def test_batch_posneg(self):
test_data = ['Worst song ever', 'Best song ever']
response = batch_sentiment(test_data, api_key=self.api_key)
self.assertTrue(isinstance(response, list))
self.assertTrue(response[0] < 0.5)
print(indicoio.sentiment('indico is so easy to use!', api_key="75b93ed62df0c77a6e58c8ebb1bb71f2", cloud="YOUR_SUBDOMAIN"))
# option 2: set module variable
indicoio.config.api_key = '75b93ed62df0c77a6e58c8ebb1bb71f2'
indicoio.config.cloud = 'YOUR_SUBDOMAIN'
print(indicoio.sentiment('indico is so easy to use!'))
import indicoio
indicoio.config.api_key = '75b93ed62df0c77a6e58c8ebb1bb71f2'
# Version >= 0.9.0
indicoio.sentiment(['indico is so easy to use!', 'Still really easy, yiss'])
# Version < 0.9.0
indicoio.batch_sentiment(['indico is so easy to use!', 'Still really easy, yiss'])
#https://indico.io/docs
#sentiment(data, [api_key], [cloud], [language])
#sentiment_hq(data, [api_key], [cloud])
#text_tags(data, [api_key], [cloud], [top_n], [threshold], [independent])
#language(data, [api_key], [cloud])
#political(data, [api_key], [cloud], [top_n], [threshold])
#keywords(data, [api_key], [version], [cloud], [top_n], [threshold], [relative]
#people(data, [api_key], [cloud], [threshold])
#places(data, [api_key], [cloud], [threshold])
#organizations(data, [api_key], [cloud], [threshold])
#twitter_engagement(data, [api_key], [cloud])
#personality(data, [api_key], [cloud])
#relevance(data, queries, [api_key], [cloud])
query.set_include_entities(False)
results = TwitterClient.search_tweets(query)
#print results['content']['statuses'][0].keys()
#tweets = [tweet['text'] for tweet in results['content']['statuses']]
tweets = []
for tweet in results['content']['statuses']:
data = {'tweet_id': tweet['id_str'], 'tweet_text': tweet['text']}
req = urllib2.Request('http://45.33.90.42:5000/insert_data')
req.add_header('Content-Type', 'application/json')
response = urllib2.urlopen(req, json.dumps(data))
tweets.append(data)
sentiment = indicoio.batch_sentiment(tweets)
pairs = sorted(zip(sentiment, tweets))
n_tweets = float(len(pairs))
top_n = 5
most_negative = pairs[:top_n]
#most_positive = list(reversed(pairs[-top_n:]))
data = {
#'most_positive': most_positive,
'most_negative': most_negative,
#'average': sum(sentiment)/n_tweets
}
pprint(zip(tweets))
