32 99-1-pb

ISSN: 2277 – 9043
International Journal of Advanced Research in Computer Science and Electronics Engineering
Volume 1, Issue 2, April 2012

Sentiment Analysis and Influence Tracking
using Twitter
Rushabh Mehta, Dhaval Mehta, Disha Chheda, Charmi Shah and Pramila M. Chawan

 share opinions on variety of topics and discuss current issues.
Abstract— An overwhelming number of consumers are Because of a free format of messages and an easy accessibility
active in social media platforms. Within these platforms of microblogging platforms, Internet users tend to shift from
consumers are sharing their true feelings about a traditional communication tools (such as traditional blogs or
particular brand/product, its features, customer service mailing lists) to microblogging services.
and how it stands the competition. With the booming of
microblogs on the Web, people have begun to express their As more and more users post about products and services they
opinions on a wide variety of topics on Twitter and other use, or express their political and religious views,
similar services. In a world where information can bias microblogging[2] web- sites become valuable sources of
public opinion it is essential to analyse the propagation and people‟s opinions and sentiments. Such data can be efficiently
influence of information in large-scale networks. Recent used for marketing or social studies. We use a dataset formed
research studying social media data to rank users by of collected messages from Twitter. Twitter contains a very
topical relevance have largely focused on the “retweet", large number of very short messages created by the users of
“following" and “mention" relations. We also perform this microblogging platform. The contents of the messages
linguistic analysis of the collected corpus and explain vary from personal thoughts to public statements.
discovered phenomena. Using the corpus, we build a
sentiment classifier, that is able to determine positive, As a microblogging and social networking website, Twitter
negative and neutral sentiments for a document. This has become very popular and has grown rapidly. An
paper discusses how Twitter data is used as a corpus for increasing number of people are willing to post their opinions
analysis by the application of sentiment analysis and a on Twitter, which is now considered a valuable online source
study of different algorithms and methods that help to for opinions. As a result, sentiment analysis on Twitter is a
track influence and impact of a particular user/brand rapid and effective way of gauging public opinion for business
active on the social network. marketing or social studies. For example, a business can
retrieve timely feedback on a new product in the market by
Index Terms—Twitter, sentiment analysis, influence, evaluating people's opinions on Twitter. As people often talk
People Rank, TwitterRank. about various entities (e.g., products, organizations, people,
etc.) in a tweet, we perform sentiment analysis at the entity
level; that is, we mine people's opinions on specific entities in
I. INTRODUCTION each tweet rather than the opinion about each whole sentence
or whole tweet. We assume that the entities are provided by
Microblogging today has become a very popular
the user, e.g., he/she is interested in opinions on iPhone (an
communication tool among Internet users. Millions of
entity).
messages are appearing daily in popular web-sites that provide
services for microblogging such as Twitter, Tumblr,
In our paper, we study how microblogging can be used for
Facebook. Authors of those messages write about their life,
sentiment analysis purposes. We show how to use Twitter as a
corpus for sentiment analysis and opinion mining. We use
 RUSHABH MEHTA Btech Computer Engineer from microblogging and more particularly Twitter for the following
VJTI,MUMBAI,INDIA
reasons:
 DHAVAL MEHTA Btech Computer Engineer from
VJTI,MUMBAI,INDIA
 DISHA CHHEDA Btech Computer Engineer from
• Microblogging platforms are used by different people to
VJTI,MUMBAI,INDIA express their opinion about different topics, thus it is a
 CHARMI SHAH Btech Computer Engineer from valuable source of people‟s opinions.
VJTI,MUMBAI, INDIA
 PRAMILA M.CHAWAN • Twitter contains an enormous number of text posts and it
Associate Professor Computer Department grows every day. The collected corpus can be arbitrarily large.
VJTI,MUMBAI,INDIA

72
All Rights Reserved © 2012 IJARCSEE

ISSN: 2277 – 9043

• Twitter‟s audience varies from regular users to celebrities, 2.”#" called the hashtag is used to mark, organize
company representatives, politicians4, and even country or alter tweets according to topics or categories.
presidents. Therefore, it is possible to collect text posts of
users from different social and interests groups. 3. “@username1" represents that a message is a reply to a user
whose user name is username1".
Sentiment is an attitude, thought, or judgment prompted by
feeling. Sentiment analysis is the process of determining and 4. Emoticons and colloquial expressions are frequently used in
measuring the tone, attitude, opinion, and emotional state of tweets, e.g. :-)", lovvve",
responses. More precisely, it is the concept of deciding lmao".
whether a specific conversation is positive, negative, or
neutral. Sentiment analysis has broad applications and 5. External Web links (e.g. http://amze.ly/8K4n0t)
encompasses work in classifying subjectivity, polarity, are also commonly found in tweets to refer to some
tonality, emotion mining, opinion mining, persuasion analysis, external sources.
and affective computing. It is a tool that allows companies to
analyze what their customers are saying regarding their 6. Length: Tweets are limited to 140 characters.
products and services, and also monitor trends in the opinions This is different from usual opinionated corpora such as
and attitudes of their customers toward the products and reviews and blogs, which are usually long.
services with respect to their competitors.
Another unique characteristic of Twitter data compared to the
There are various types of marketing strategies such as mass other opinionated corpora is its volume. It is estimated that
marketing, segmentation and one to one marketing. One to one people post about 60 million tweets every day and the number
marketing is an effort to find individual customer's needs and is still increasing rapidly.
to provide a good response for them. Recommender systems
have appeared in e-commerce problems to support product
recommendation, which provide one to one marketing. Indeed, II. METHODOLOGY
recommender systems individualize the way of recommending
products. These systems try to recommend different products
to each customer with collecting data of customer preferences
A. Data Collection
and data mining techniques. Recommender systems have
Twitter has an open API that allows anyone to get a list of a
recently become popular among many well-known e-
user's friends (provided the account is not private) It is
businesses such as Amazon.com, MovieFinder.com.
therefore easy to create a graph of the network. Since there are
more than 100M nodes in this graph with many times that
As people often talk about various entities (e.g., products,
many edges, it requires a lot of computational power to
organizations, people, etc.) in a tweet, we perform sentiment
process this entire graph. I therefore propose to focus on a
analysis at the entity level; that is, we mine people's opinions
smaller subset. However, recently Twitter has been more
on specific entities in each tweet rather than the opinion about
circumspect in allowing unfettered access to the entire social
each whole sentence or whole tweet. We assume that the
graph and tweet stream. It allows this access termed the "fire
entities are provided by the user, e.g., he/she is interested in
hose" to a small chosen set of companies only. Through the
opinions on iPhone (an entity). One approach to perform
public API, one can only access a single user's tweet stream
sentiment analysis is based on a function of opinion words in
and his profile information and also the public timeline of
context. Opinion words are words that are commonly used to
tweets.
express positive or negative sentiments, e.g., good" and bad".
The approach generally uses a dictionary of opinion words to
The Streaming API is the real-time sample of the Twitter
identify and determine sentiment orientation (positive,
Firehose. This API is for those developers with data intensive
negative or neutral). The dictionary is called the opinion
needs. If you're looking to build a data mining product or are
lexicon.
interested in analytics research, the Streaming API is most
suited for such things. Streaming API allows for large
quantities of keywords to be specified and tracked, retrieving
geo-tagged tweets from a certain region, or have the public
Twitter Data
statuses of a user set returned. This requires you to establish a
long-lived HTTP connection and maintain that connection.
Twitter has developed its own language conventions.The
following are examples of Twitter conventions.
The Twitter Search API is a dedicated API for running
1. “RT" is an acronym for retweet, which is put in
searches against the real-time index of recent Tweets. If you're
front of a tweet to indicate that the user is repeating
currently developing on the Search API, and find that your
or reposting.
application is being rate-limited or you just have aggressive

73

ISSN: 2277 – 9043

querying needs, then you should be moving over to the
Streaming API. Classifications

Four classifications are used in this corpus:
B. Analysis
Using Twitter API we collected a corpus of text posts and Positive ● Positive indicator on topic
formed a dataset of three classes: positive sentiments, negative Neutral ● Neither positive nor negative
sentiments, and a set of objective texts (no sentiments). We indicators
queried Twitter for two types of emoticons: ● Mixed positive and negative indicators
• Happy emoticons: “:-)”, “:)”, “=)”, “:D” etc. ● On topic, but indicator undeterminable
• Sad emoticons: “:-(”, “:(”, “=(”, “;(” etc. ● Simple factual statements
● Questions with no strong emotions
The two types of collected corpora will be used to train a indicated
classifier to recognize positive and negative sentiments.
Because each message cannot exceed 140 characters by the Negative ● Negative indicator on topic
rules of the microblogging platform, it is usually composed of Irrelevant ● Not English language
a single sentence. Therefore, we assume that an emoticon ● Not on-topic (e.g. spam)
within a message represents an emotion for the whole message
and all the words of the message are related to this emotion.

Sentiment assignment is an extremely subjective exercise.
i. Preparing the Data
The data set contains information on X million profiles. For this corpus, “Positive” and “Negative” labels were
Profile information is limited to the user accounts followed by reserved for tweets which clearly express an emotion or where
the user. Since the data set is as of a certain date, the the implications were unambiguous. As a rule of thumb,
information is not complete as of today. However, the data set “neutral” was the preferred label for border line cases.
contained enough information to create training and test data
sets. Examples:

All the tweets would be initially considered as a bag of words. There are huge lines at the @apple store.
Labeled neutral. From a shoppers perspective this could be
for eg. "This is excellent" bad, or it could be a sign of excitement about the product
launch. From an investor‟s perspective this could be good,
would not be considered as a string but as a bag of three words since it indicates a strong new product launch.
"This", "is" and "excellent".
I had to wait for six friggin’ hours in line at the @apple
Then the stop words such as "the", "a", "with" etc will be store.
removed from the bag as these words do not have any Labeled negative. The tweeter is clearly unhappy with the
sentiment expressing nature. Once these non-sentimental stop situation and is referring to Apple in the negative sense.
words are are removed and hence the corpus refined, the
process of sentiment analysis can begin.
iii. Preprocessing
Data preprocessing consists of three steps: 1) tokenization,
ii. Sentiment gradation 2) normalization, and 3) part-of-speech (POS) tagging.[12]
The bag of sentiment expressive words i.e. every tweet is now
analyzed in parts. A knowledge base is created which has the Emoticons and abbreviations (e.g., OMG, WTF, BRB) are
relative sentiments of words denoted by a floating point identified as part of the tokenization process and treated as
number ranging from -1 to 1. individual tokens.

All the words in the bag are cross checked across this For the normalization process, the presence
knowledge base. This gives the sentiment of ever word in the of abbreviations within a tweet is noted and then abbreviations
range. After this, taking into consideration the type of words are replaced by their actual meaning (e.g., BRB
and their sentiment score, the sentiment of the overall tweet is - > be right back). We also identify informal intensifiers
calculated. This would determine what the sentiment of the such as all-caps (e.g., I LOVE this show!!! and character
tweet is and how the user has expressed his satisfaction over repetitions (e.g., I‟ve got a mortgage!! happyyyyyy”), note
the product or service. their presence in the tweet. All-caps words are made into
lowercase, and instances of repeated charaters are replaced by
a single character.
74

ISSN: 2277 – 9043

Finally, the presence of any special Twitter b. Usage of links. Users very often include links in their
tokens is noted (e.g., #hashtags, usertags, and URLs) and tweets. An equivalence class was created for all URLs. That
placeholders indicating the token type are substituted. Our is, a URL like "http://tinyurl.com/cvvg9a" was converted to
hope is that this normalization improves the performance of the symbol "URL."
the POS tagger, which is the last preprocessing step.
c. Usernames. Users often include usernames in their tweets,
in order to address messages to particular users. A de facto
iv. Feature-based extraction standard is to include the @ symbol before the username (e.g.
@alecmgo). An equivalence class was made for all words that
The collected dataset is used to extract features that will be started with the @ symbol.
used to train our sentiment classifier. We used the presence of
an n-gram as a binary feature, while for general information d. Removing the query term. Query terms were stripped out
retrieval purposes, the frequency of a keyword‟s occurrence is from Tweets, to avoid having
a more suitable feature, since the overall sentiment may not the query term affect the classification.
necessarily be indicated through the repeated use of keywords.
A. Process of constructing n-grams
2.Bigrams
The reason we experimented with bigrams was we wanted to
1. Filtering – we remove URL links (e.g. smooth out instances like 'not good' or 'not bad'. When
http://example.com), Twitter user names (e.g. negation as an explicit feature didn't help, we thought of
@alex – with symbol @ indicating a user name), experimenting with bigrams.
Twitter special words (such as “RT”), and emoticons.

2. Tokenization – we segment text by splitting it by B. Negate as a features
spaces and punctuation marks, and form a bag of NEGATE is added as a specific feature which is added when
words. However, we make sure that short forms such “not” or „n‟t” are observed in the dataset. [7]
as “don‟t”, “I‟ll”, “she‟d” will remain as one word.
C. Part of Speech (POS) features
3. Removing stopwords – we remove articles (“a”, “an”, We felt like POS tags would be a useful feature since how you
“the”) from the bag of words. made use of a particular word. For example, „over‟ as a verb
has a negative connotation whereas „over‟ as the noun, would
4. Constructing n-grams – we make a set of n-grams out refer to the cricket over which by itself doesn‟t carry any
of consecutive words. A negation (such as “no” and negative or positive connotation.
“not”) is attached to a word which precedes it or follows
it. For example, a sentence “I do not like fish” will form D. Lexicon features
two bigrams: “I do+not”, “do+not like”, “not+like fish”. Words listed the MPQA subjectivity lexicon (Wilson,
Such a procedure allows to improve the accuracy of the Wiebe, and Hoffmann 2009) are tagged with their prior
classification since the negation plays a special role in an polarity:positive, negative, or neutral.We create three features
opinion and sentiment expression. based on the presence of any words from the lexicon.

1.Unigram iv. Literature Review on taggers
Building the unigram model took special care because the The models included for sentiment analysis in our paper can
Twitter language model is very different from other domains be downloaded for the POS tagger website at
from past research. The unigram feature extractor addressed http://nlp.stanford.edu/software/tagger.shtml . All taggers are
the following issues: accompanied by the props files used to create them,given
below is a more detailed information about the creation of the
a. Tweets contain very casual language. For example, you taggers.
can search "hungry" with a random number of u's in the
middle of the word on http://search.twitter.com to understand For English, the bidirectional taggers are slightly more
this. Here is an example sampling: accurate, but tag much more slowly; choose the appropriate
huuuungry: 17 results in the last day tagger based on your speed/performance needs.
huuuuuuungry: 4 results in the last day
huuuuuuuuuungry: 1 result in the last day English taggers
Besides showing that people are hungry, this emphasizes the ---------------------------
casual nature of Twitter and wsj-0-18-bidirectional-distsim.tagger
the disregard for correct spelling.

75

ISSN: 2277 – 9043

Trained on WSJ sections 0-18 using a bidirectional construct some implicit signals from the user's content stream
architecture and including word shape and distributional that are analogous to recommendation. Specifically, I look at
similarity features. three signals that are counted as up votes. First, if a user
Penn Treebank tagset. follows another account, that is considered a positive rating
Performance: for the account that is followed. Second, if a user retweets (i.e.
97.28% correct on WSJ 19-21 echoes a tweet to his own tweet stream), that can also
(90.46% correct on unknown words) considered a positive rating. Thirdly, if a user shares a
"hashtag" with another user, that is considered a positive
wsj-0-18-left3words.tagger rating for the user who is being followed. Sharing a hashtag
Trained on WSJ sections 0-18 using the left3words implies that the two tweets are related to the same topic,
architecture and includes word shape features. Penn tagset. although they may express two entirely different opinions (for
Performance: e.g. the recent controversy around wikileaks elicited a storm of
96.97% correct on WSJ 19-21 either vehement approval or disapproval from twitter users,
(88.85% correct on unknown words) but they used the same #wikileaks hashtag).

wsj-0-18-left3words-distsim.tagger
Trained on WSJ sections 0-18 using the left3words
architecture and includes word shape and distributional
similarity features. Penn tagset. III.ALGORITHMS
Performance:
97.01% correct on WSJ 19-21
(89.81% correct on unknown words) B. PeopleRank Algorithm
In general, global knowledge of network topology can make
english-left3words-distsim.tagger for very efficient routing and forwarding decisions.
Trained on WSJ sections 0-18 and extra parser training data Collecting and exchanging topology information in
using the left3words architecture and includes word shape and opportunistic networks is cumbersome because of their
distributional similarity features. Penn tagset. intermittent connectivity and unpredictable mobility.

english-bidirectional-distsim.tagger PeopleRank is inspired by the PageRank [5] algorithm
Trained on WSJ sections 0-18 using a bidirectional employed by Google to rank web pages. By crawling the
architecture and including word shape and distributional entire web, this algorithm measures the relative importance
similarity features. of a page within a graph (web). Motivated by the success of
Penn Treebank tagset. this algorithm, we propose to apply a similar technique,
which we call PeopleRank to rank the nodes in a social
wsj-0-18-caseless-left3words-distsim.tagger graph. The main idea is that nodes with a higher
Trained on WSJ sections 0-18 left3words architecture and PeopleRank value will generally be more “central” in the
includes word shape and distributional similarity features. social graph.
Penn tagset. Ignores case.
a. Centralized Peoplerank
english-caseless-left3words-distsim.tagger In PeopleRank we tag people as “important” when they are
Trained on WSJ sections 0-18 and extra parser training data linked (in a social context) to many other “important”
using the left3words architecture and includes word shape and people. We assume that only neighbors in the social graph
distributional similarity features. Penn tagset. Ignores case. have an impact of the popularity.

a social graph Gs = (Vs,Es) as a finite undirected graph with
v. Inferring Edge Strength a vertex set V and an edge set Es. An edge (u, v) ∈ Es if,
In the simplest setting, a user being connected to another user and only if, there is a social relation between nodes u and v.
can be used as a preference signal. In recent times, given the In this paper, we define a social relationship between two
explosive growth of twitter, there have emerged a large nodes u and v either (i) if they are declared friends, or (ii) if
number of "bot" accounts that seek to follow as many users as they are sharing k common interests.
possible in the hope that unwitting users will follow them
back. Therefore, looking at "followed" accounts yields more
information about the account holder's preferences rather than b. Distributed PeopleRank
"follower" accounts. The distributed version of PeopleRank is shown in
Algorithm. In this version, whenever two neighbor nodes in
In a traditional item recommendation setting, users rate items the social graph meet, they exchange two pieces of
on a scale of 1-5 or by an up or down vote. In twitter, there is information:
no explicit rating of accounts by other users. However, we can

76

ISSN: 2277 – 9043

others repeat your message either through literal retweets or
(i) their current PeopleRank values; and more subtle gestures, such as replies and repeating the URLs
(ii) the number of social graph neighbors they have. that you tweet. If someone on Twitter receives your message
Then, the two neighbors update their PeopleRank through a trusted intermediary, it is assigned a much greater
values level of trust. So the goal is to get influential people to follow
you and then act as a conduit for your marketing.[10]

Influence is a sophisticated measure of a user‟s relative
importance among the entire Twitter network.
C. TwitterRank
Uses various statistics about a handle as parameters like
TwitterRank measures the influence taking both the topical number of followers, retweets, mentions, URL‟s shared.
similarity between users and the link structure into account. There are three major components that add up to the score:
In a dataset prepared for this study, it is observed that 1)72.4% your followers, your mentions and retweets, and your lists, all
of the users follow more than 80% of their followers, and (2) accounted as ratios between you and others.
80.5% of the user have 80% of their friends follow them
back.[4]Our study reveals that the presence of “reciprocity” can Followers is the strongest component of the calculation is the
be explained by phenomenon of homophily.Based on this number of followers you have. In my opinion, your presence
finding, TwitterRank, an extension of PageRank algorithm, is on Twitter and getting followers can be influenced by at least
proposed to measure the influence of users in Twitter. the following three major factors concerning you and your
TwitterRank measures the influence taking both the topical Twitter account:
similarity between users and the link structure into account.
Experimental results show that TwiterRank outperforms the i. Persona – how known you are. Measured by the
one Twitter currently uses and other related algorithms, number of followers you have, our time on
including the original PageRank and Topic-sensitive PageRank. Twitter.
ii. Engagement – how engaged you are. Measured by the
number of followers you have, compared the
First, it potentially brings order to the real-time web in that it number of people you follow; Measured by the
allows the search results to be sorted by the authority/influence number of followers you have, compared to the
of the contributing twitterers giving a timely update of the number of mentions and retweets you‟ve made.
thoughts of influential twitterers. Second, Twitter is also a iii. Wits – how smart and creative your tweets are.
marketing platform. Targeting those influential users will Measured by the number of followers you have
increase the efficiency of the compared to the total number of tweets you've
marketing campaign. For example, a handphone manufacturer made.
can engage those twitterers influential in topics about IT
gadgets to potentially influence more people . There are also For this part, the followers/following ratio the weight of 3, the
applications that utilize Twitter to gather opinions and followers/tweets a weight of 2 and the followers/time a weight
information on particular topics. Identifying influential of 1. The followers/(mentions + retweets) has a weight of 0.5
twitterers for interesting topics can improve the quality of and works in the negative way, so people who bother other
opinions gathered. people get a bit of a minus to their followers result. Besides,
those who are able to get the same number of followers
PageRank improves over in-degree by considering the link without mentioning people, must have a small advantage.
structure of the whole network. Nevertheless, Pagerank
ignores the interests of twitterers, which affects the way
twitterers influence one another. Our proposed approach The second most important part of the calculation is the ratio
addresses the shortcomings of in-degree and PageRank by between mentions and being mentioned, together with the
taking into account both the link structure and topical number of retweets you get with the absolute "reach" of those
similarity among twitterers. retweets (measured in the number of people who follow
people that retweeted you). A similar reach is also accounted
In the context of Twitter, homophily in the mentions and replies.
implies that a twitterer follows a friend because she is
interested in some topics the friend is publishing, and the Twitter lists are getting used more and more, so they are also
friend follows back because she finds they share similar considered in the calculation. The number of lists you appear
topical interest. on, the number of people who follow those lists and the
number of people, who follow lists you've created are the
C. Influence Tracking basic parameters for the calculation. This component adds
only a small bit to the final score.
In many ways Twitter-based marketing is like a pyramid
scheme. While sending tweets to your own followers is one
way of broadcasting a message, it is more effective to have
77

ISSN: 2277 – 9043

The three major components currently have the following products, services, and spark new waves of tweets gushing
weight in the final score: with positive sentiment. Doing so over time helps to build
the social, and more relevant, business of the future while
 Followers: around 60% improving relationships to convert followers into
 Mentions and retweets: around 30% stakeholders.
 Lists: around 10%

D. Model for calculating influence REFERENCES
[1] Bo Pang and Lillian Lee, “Opinion Mining and Sentiment Analysis”,
The assumptions about the model: Foundations and Trends in Information Retrieval
Vol. 2, No 1-2 (2008)
1. Influence(X) = Expected number of people who will read a [2] Alexander Pak, Patrick Paroubek, “Twitter as a Corpus for Sentiment
Analysis and Opinion Mining”
tweet that X tweets, including all retweets of that tweet. For [3] Aditya Pal & Scott Counts, “Identifying Topical Authorities in
simplicity, we assume that, if a person reads the same message Microblogs”, WSDM‟11, February 9–12, 2011, Hong Kong, China,
twice (because of retweets), both readings count. Copyright 2011 ACM
[4] Jianshu Weng, Ee-Peng Lim, Jing Jiang, Qi He, “TwitterRank: Finding
Topic-sensitive Influential Twitterers”, WSDM‟10, February 4–6, 2010,
2. If X is a member of Followers(Y), then there is a New York City, New York, USA Copyright 2010 ACM
1/||Following(X)|| probability that X will read a tweet posted [5] Abderrahmen ,Mtibaa Martin May Christophe Diot Mostafa Ammar,
by Y, where Following(X) is the set of people that X follows. “PeopleRank: Social Opportunistic Forwarding”
[6] Albert Bifet and Eibe Frank,” Sentiment Knowledge Discovery in
Twitter Streaming Data”
3. If X reads a tweet from Y, there‟s a constant probability p [7] Alec Go , Lei Huang and Richa Bhayani,” Twitter Sentiment Analysis”,
that X will retweet it. CS224N - Final Project Report June 6, 2009.
[8] B. Jansen, M. Zhang, K. Sobel, A. Chowdury. The Commerical Impact of
This model is obviously simplistic in all three assumptions. Social Mediating Technologies: Micro-blogging as Online Word-of-Mouth
Branding, 2009.
But it‟s a reasonable first cut. In particular, it accounts for the [9] C. Manning and H. Schuetze. Foundations of Statistical Natural Language
inflation that occurs from people who follow in the hopes of Processing,1999.
reciprocity. There‟s less value in being followed by someone [10] D. Kempe, J. Kleinberg, and E. Tardos., “Maximizing
who follows a lot of people, because that person is less likely the spread of influence through a social network”, In KDD ‟03: Proceedings
of the ninth ACM SIGKDD international conference on Knowledge discovery
to read your messages or retweet them. and data mining, pages 137–146, New York, NY, USA, 2003. ACM.
[11] W. Zhang and S. Skiena., “Improving movie gross prediction through
Of course, there‟s room for adding more realism to this model, news analysis”, In Web Intelligence, pages 301304, 2009.
but it is at least close enough to the truth to be interesting. [12] Efthymios Kouloumpis, TheresaWilson, Johanna Moore “Twitter
Sentiment Analysis:The Good, the Bad and the OMG”, Proceedings of the
Fifth International AAAI Conference on Weblogs and Social Media
From this model, it‟s easy to measure someone‟s influence
recursively, assuming that we know the constant retweet
probability p:

Influence(X) = ∑ (1+p * Influence(Y))/||Following(Y)
Followers(X)

IV. CONCLUSION
Microblogging nowadays became one of the major types of
the communication. A recent research has identified it as • RUSHABH MEHTA
online word-of-mouth branding.The era of analysis
paralysis is officially over. Instead of just listening,
companies can now study people and their interests based
on what they say and do and also how they color their
profiles.
Rushabh Mehta is a Final Year B.Tech student of Computer Technology at
This goldmine of insight gives brands the potential to VJTI. He gave his HSC from Ramnivas Ruia College securing 93.83% &
improve marketing, promotional and advertising campaigns stood 45th out of 2,20,000 students in Engineering Entrance Exam. Currently
his CGPA is 9.1 at VJTI. He has pursued internships at Cisco & IIT-Bombay.
to start. As this practice develops, brands can also gather the Being the technology evangelist, he has co-founded CSI chapter of VJTI as
intelligence necessary, and widely available, to improve well.
78

ISSN: 2277 – 9043

publications to her credit. She has guides 35 M. Tech. projects & 85 B. Tech.
projects.

Some publications:
• DHAVAL MEHTA
1. Paper on „Grid FTP Protocol combined with Data Grid use for sharing
files‟ – Published in „ETCC-08 - National Conference on emerging trends in
Computing & Communication‟ at NIT, Hamirpur (30-31 Dec 2008)

2. Better Approach to Requirement Engineering with Agile Process Nirmala
Shinde, Mansi U Kulkarni, Pramila Chawan
Published in RTICSIT- National Conference On Recent Trends In Computer
Science & Information Technology at Guru Nanak Dev
Engineering College, Mailoor Road Bidar(9-10 May 2009)
Dhaval Mehta is a Final Year B.Tech student of Computer Technology at
VJTI. He gave his HSC from KC College securing 95% & stood 22nd out of 3. Archana S. Sumant & Pramila M. Chawan, Smart Cards & Biometrics
2,20,000 students in Engineering Entrance Exam. With 197/200.Currently his :Integration Of Two Growing Technologies, International Conference &
CGPA is 8.7 at VJTI. He has co-founded CSI chapter of VJTI as well. Workshop on Emerging Trends in Technology 2010 (ICWET 2010), ISBN
978-1-60558-812-4

• DISHA CHHEDA 5. Mrs. Pramila M. Chawan Mr. Sandip Shingade Mr. Pravin Bansode,
Retrieving images on World Wide Web, The 2nd National Conference On
Recent Trends in Computer Engineering (RTCE 2009)

6. Ajinkya Patil, Apurva Mayekar, Shruti Gurye, Varun Karandikar and
Pramila M. Chawan, Audio Streaming on mobile phones, International
Journal of Science and Engineering Research 2011,IJSER-11, June 2011.

7. Deepali kadam, Nandan Bhalwankar, Rahul Neware, Rajesh Sapkale,
Raunika lamage and Pramila M. Chawan, Oracle Real Application Clusters,
Disha Chheda is a Final Year BTech Student of Computer Engineering at
International Journal of Science and Engineering Research 2011,IJSER-11,
VJTI. She has completed her Diploma in Computer Technology from
June 2011
Vivekanand Education Society’s Polytechnic, Chembur in 2009 and was a
topper in Mumbai division of MSBTE with aggregate of 92% marks. She has
been studying in VJTI since 2009 and will be graduating in 2012. Her
Cumulative Performance Index is 8.7/10. She has participated in many
college-level academic and extra-curricular competitions and has even had
an experience in managing events in different college level festivals.

• CHARMI SHAH

CHARMI SHAH has done diploma in computer engg from
K.J.SOMAIYA POLYTECHNIC having scored 91.38% and right
now pursuing degree from V.J.T.I . She is very hard working, easy to
grasp things and can easily adopt any new environment. She has previously
worked with vb.net and java language for project purposes.

• PRAMILA M.CHAWAN

Pramila M. Chawan is currently working as an Associate
Professor in the Computer Technology Department of
“Veermata Jijabai Technological Institute (V.J.T.I.),
Matunga, Mumbai (INDIA)”. She received her
Bachelors’ Degree in Computer Engineering from
V.J.T.I., Mumbai University (INDIA) in 1991 & Masters’
Degree in Computer Engineering from V.J.T.I., Mumbai
University (INDIA) in 1997.She has an academic
experience of 20 years. She has taught Computer related subjects at both
Undergraduate & Post Graduate levels. Her areas of interest are Software
Engineering, Software Project Management, Management Information
Systems, Advanced Computer Architecture & Operating Systems. She has
published 12 papers in National Conferences and 7 papers in International
Conferences & Symposiums. She also has 16 International Journal

79

32 99-1-pb

More Related Content

What's hot (20)

Similar to 32 99-1-pb (20)

More from Mahendra Sisodia (15)

Recently uploaded (20)

32 99-1-pb