International Journal of Engineering and Technical Research (IJETR)

ISSN: 2321-0869 (O) 2454-4698 (P), Volume-5, Issue-1, May 2016

Depicting the Public Sentiment Variations on Twitter

Shubham Pacharne, Vaibhav Sonawane, Sahil Rajeshirke, Pranav Kolhatkar

Hashtags are words prefixed with # and are used to

Abstract Twitter platform is valuable to follow the public indicate the topics of tweets. For example, #Election2014
sentiments. Knowing users point of views and reasons behind can be used in tweets related to Indias General Election of
them at various point is an important study to take certain
decisions. Categorization of positive and negative opinions is a
2014. Hashtags play an important role in Twitter. Popular
process of sentiment analysis. It is very useful for people to nd
sentiment about the person, product etc. before they actually
hashtags can become trending topics in the home page of
make opinion about them. In this project, we stream tweets Twitter.
based on a topic and then plot a pie chart, which represents the TABLE I
percentage of positive and negative sentiments in a convenient
manner. We also display the number of tweets on which the EXAMPLES OF TWITTER POSTS WITH USERS OPINIONS
result is based. Instead of analysing individual sentiment, we
stream multiple sentiments and represent percentage of positive Shubham Pacharne: Python is the best programming
and negative sentiments about that topic.
language. http://bit.ly/1Ro76HT
Sundar Pichai: Tremendous excitement among Googlers
Index Terms Sentiment analysis, Tokenization, hashtags,
Stop words, Part-of-speech tagging, NLTK, Nave-Bayes for PM Modi's visit. #ModiInUSA
classifier Darklightnjh: I hate how when im watching a YouTube
video the ad is all hq but when the real video starts its DSI
I. INTRODUCTION camera quality.
Semantria: "I love the summer in New York, but I hate the
Twitter today has become a very popular communication
winter."
tool among web savvies. Millions of tweets are appearing
daily. Authors of these messages write about their life, Curiosity Rover: Namaste, @MarsOrbiter!
share opinions on variety of topics and discuss current issues. Congratulations to @ISRO and India's first interplanetary
As more and more users post about products and services mission upon achieving Mars orbit.
they use, or express their political and religious views,
micro-blogging web-sites become valuable sources of II. LITERATURE SURVEY
peoples opinions and sentiments. Such data can be efficiently Sentiment analysis is the process of analysing the opinions
used for marketing, social studies or improving services. which are extracted from different sources like the comments
Twitter contains a very large number of very short messages given on forums, reviews about products, various policies
(up to 140 characters) created by the users. and the topics mostly associated with social networking sites
The contents of the messages vary from personal thoughts and tweets. A very broad overview of the existing work was
to public statements. Table 1 shows examples of typical presented in (Pang and Lee, 2008). In their survey, the authors
posts from Twitter. As the audience of micro-blogging describe existing techniques and approaches for an opinion
platforms and services grows every day, data from these oriented information retrieval. However, not many researches
sources can be used in sentiment analysis tasks. For example, in opinion mining considered blogs and even much less
restaurants may be interested in the following questions: addressed micro-blogging. In (Yang et al., 2007), the authors
use web-blogs to construct corpora for sentiment analysis.
What do people think about us (food, service etc.)? The authors applied SVM and CRF learners to classify
How positive (or negative) are people about our sentiments at the sentence level and then investigated several
food-items? strategies to determine the overall sentiment of the document.
As the result, the winning strategy is defined by considering
the sentiment of the last sentence of the document
Political parties may be interested to know if people support
as the sentiment at the document level.
their program or not. News channels may ask peoples
opinion on current debates (News hour).
In (Go et al., 2009), authors used Twitter to collect training
data and then to perform a sentiment search. The approach is
Shubham Pacharne, Department of Computer Engineering, P.E.Ss similar to (Read, 2005). The authors construct corpora by
Modern College of Engineering, Shivaji Nagar, Pune 411005, India. using emoticons to obtain positive and negative samples,
and then use various classifiers. The best result was obtained
Vaibhav Sonawane, Department of Computer Engineering, P.E.Ss by the Naive Bayes classifier with a mutual information
Modern College of Engineering, Shivaji Nagar, Pune 411005, India.
measure for feature selection. The authors were able to obtain
Sahil Rajeshirke, Department of Computer Engineering, P.E.Ss Modern up to 81% of accuracy on their test set. However, the method
College of Engineering, Shivaji Nagar, Pune 411005, India. showed a bad performance with three classes (negative,
positive and neutral).
Pranav Kolhatkar, Department of Computer Engineering, P.E.Ss
Modern College of Engineering, Shivaji Nagar, Pune 411005, India

23 www.erpublication.org
 Depicting the Public Sentiment Variations on Twitter

III. ARCHITECTURE DIAGRAM

The training set is trained against by the Nave Bayes

IV. SENTIMENT TRACKING classifier and the testing set is tested against for accuracy.
We have 2 datasets one containing 5000+ positive When user enters topic, streaming tweets are fetched from
sentences and other containing 5000+ negative sentences the Twitter API about that topic.

Pre-processing tasks (tokenization, stop word removal, Each tweet is tokenized and a list is created which contains
part-of-speech tagging) are done using nltk on each sentence the words followed by the Boolean value indicating whether
of the datasets. or not the words are present in the word_features list.
Using the Nave Bayes classifier, the tweet is classified as
(1) Tokenization We segment text by splitting it by spaces positive or negative, depending on whether the words in the
and punctuation marks, and form a bag of words. tweet appear more frequently in the positive category or the
(2) Stop word removal Stop words are natural negative category.
language words which have very little meaning, such as
"and", "the", "a", "an", and similar words. A Pie Chart is generated showing the percentage of
(3) Part-of-speech tagging It is a sentence-based process and positive and negative tweets about that topic out of the total
given a sentence formed of a sequence of words, number of tweets.
part-of-speech tagging tries to label (tag) each word with its
correct part-of-speech. The number of tweets based on which the pie chart is
Therefore, pre-processing techniques on tweets are necessary generated is also displayed.
for obtaining satisfactory results on sentiment analysis

We create a word_features list from which only the V. RESULTS

features which appear most frequently are extracted. In this section, we can segregate the reviews based on
different categories and generate graphical representation of
This list is used to create feature set list which contains all the sentiments in the form of pie-charts.
the words from the two datasets and a Boolean value The user will enter the topic and press the button Get
indicating whether or not that particular word is from the most Streaming Twitter Data. The software will start getting the
commonly occurring words list. streaming Twitter data based on topic from the Twitter API.
Then user will click on Generate Pie Chart.
This feature set randomly shuffled and split into 2 sets-
majority of it in training set and remaining in
testing set.

24 www.erpublication.org
 International Journal of Engineering and Technical Research (IJETR)
ISSN: 2321-0869 (O) 2454-4698 (P), Volume-5, Issue-1, May 2016
1. Pie-Chart representation ACKNOWLEDGMENT
Software will generate a pie-chart showing the percentage of We take this opportunity to express our profound gratitude
positive and negative sentiments based on the users entered and deep regards to our mentor Mr.Pradeep Pattayat (QE
topic. Lead, Persistent Systems) for his exemplary guidance,
monitoring and constant encouragement throughout the
course of this thesis.

We also take this opportunity to express a deep sense of

gratitude to Dr. B. D. Phulpagar, Mr. Vijeth Rao for their
cordial support, valuable information and guidance, which
helped us in completing this task through various stages.

REFERENCES
[1] Alexander Pak, Patrick Paroubek, Twitter as a Corpus for Sentiment
Analysis and Opinion Mining, F-91405 Orsay Cedex, France.
[2] D. Chakrabarti and K. Punera. Event summarization using tweets. In
Proc. of the Fifth International AAAI Conference on Weblogs and
Social Media, Barcelona, Catalonia, Spain, 2011.
[3] B. Pang and L. Lee, Opinion mining and sentiment analysis, Found.
Trends Inform. Retrieval, vol. 2, no. (12), pp. 1135, 2008.
[4] M. Hu and B. Liu, Mining and summarizing customer reviews, in
Proc. 10th ACM SIGKDD, Washington, DC, USA, 2004.
2. Number of tweets based on which Pie-Chart is plotted [5] D. Tao, X. Tang, X. Li, and X. Wu,Asymmetric bagging and random
subspace for support vector machines-based relevance feedback in
image retrieval, IEEE Trans. Patt.Anal. Mach. Intell., vol. 28, no. 7,
pp.10881099, Jul. 2006
[6] O. Tsur and A. Rappoport, Whats in a hashtag?: content based
prediction of the spread of ideas in microblogging communities, in
WSDM, 2012,
pp. 643652.
[7] A. Cui, M. Zhang, Y. Liu, S. Ma, and K. Zhang, Discover breaking
events with popular hashtags in twitter, in CIKM, 2012,
pp.1794-1798.
[8] H. Kwak, C. Lee, H. Park and S. B. Moon, What is twitter, a social
network or a news media? in WWW, 2010, pp 591-600

VI. CONCLUSIONS
Nowadays, the opinions and reviews of people on social
networking sites like Twitter are hugely inuential on other
people and their decisions. This system can help such people
nd about sentiments about a topic on Twitter and make
necessary decisions conveniently.

Despite all the challenges and potential problems that

threatens Sentiment analysis, one cannot ignore the value that
it adds to the industry. Because Sentiment analysis bases its
results on factors that are so inherently humane, it is bound to
become one the major drivers of many business decisions in
future. Improved accuracy and consistency in text mining
techniques can help overcome some current problems faced in
Sentiment analysis.

Looking ahead, what we can see is a true social democracy

that will be created using Sentiment analysis, where we can
harness the wisdom of the crowd rather than a select few
experts. A democracy where every opinion counts and every
sentiment affects decision making.

25 www.erpublication.org