International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169

Volume: 2 Issue: 8 2094 – 2097

_______________________________________________________________________________________________

Emergent System using Tweet Analyzer: Naturally Inspired Computing

Approach
N.sharmila1, B .Anantharaj2, R.Hariraman3
Student1, HOD2, Faculty3 Department of CSE,
Thiruvalluvar college of Engineering and Technology
Anna University, India
Email: [email protected] 1/ [email protected]/ [email protected] 3

Abstract--Nowadays much system has developed to reach the people during disasters. A social interaction with the micro blogging services has
vastly increased. Twitter a well popular social medium has scarcity of attention makes people interacting with each other. This paper explores
the use of twitter for disaster event using text classifier. It analyses the targeted event with tweeted text and identifies the target and its location.
By the use of SVMs, the text classifiers are achieved.SVM performs the methods of Bayesian filtering for the informative messages of particular
event. This method of filtering works best compare with the other methods for estimating the informative messages. As an application we are
sending an E-mail and SMS alert message through the twitter and the declared set of friends and followers follows the messages.

Index terms - Twitter, Informative, SVM, Bayesian Filtering.

__________________________________________________*****_________________________________________________

1. INTRODUCTION the world. Twitter is an online social networking and micro

blogging service that enables users to send and read
1.1 Naturally Inspired Algorithm.
"tweets", which are text messages limited to 140 characters.
Nature is the best teacher and its designs and Registered users can read and post tweets but unregistered
capabilities are extremely enormous and mysterious that users can only read them. It's a platform wherein users share
researchers are trying to mimic nature in technology. Bio their thoughts, news, information and jokes. Twitter makes
inspired computing has come up as a new era in computing global communication cheap and measurable. Profiles are
encompassing a wide range of applications, covering all public anyone in the world can see what you write, unless
most all areas including computer networks, security, you elect to make your profile private. Users "follow" each
robotics, bio medical engineering, control systems ,parallel other in order to keep tabs on and converse with specific
processing ,data mining, power systems, production people
engineering and many more. New or persistent problems in
computer science could have a lot in common with problems Twitter users are able to post straight and derivative
nature has encountered and resolved long ago. Thus an easy updates. Straight posts are used when a user aims her update
mapping is possible between nature and technology. to a specific person, whereas derivative updates are used
when the update is meant for anyone that cares to read it.
Towards technology through Nature Even though straight updates are used to communicate
directly with a specific person, they are public and anyone
The real beauty of nature inspired algorithms lies in can see them. Often times two or more users will have
the fact that it receives its sole inspiration from nature. They conversations by posting updates directed to each other.
have the ability to describe and resolve complex Around 25.4% of all posts are directed, which shows that
relationships from intrinsically very simple initial conditions this feature is widely used among Twitter users.
and rules with little or no knowledge of the search space A predominant characteristic that is common
Nature is the perfect example for optimization, because if among the micro blogging services is their real-time nature.
we closely examine each and every features or phenomenon Although blog users usually update their web logs once
in nature it always find the optimal strategy, still addressing every days, twitter user usually write tweets many times in a
complex interaction among organisms ranging from day. Several important occurrences symbolize their real time
microorganism to fully fledged human beings, balancing the nature.
ecosystem, maintaining diversity, adaptation, physical Many people have made research work on the
phenomenon like river formation, forest fire ,cloud, rain .etc social media, Twitter most of the studies based on the
structure of twitter [2],[3],[4],and some of them on the
1.2 Twitter characteristic of twitter[5],[6] and some on the application
usage of twitter[7][8].
Social media, a very old and pervasive mechanism This paper illustrates the detection of the
for communicating distal interactions among people, have earthquake disaster by monitoring the tweet messages and it
become widespread in the age of Web[1]. With interfaces helps to detect the occurrence in the real –time nature.
that allow people to follow the live of friends, social contact This paper presents the probe of the real-time
and family. This network is used millions of people around nature. Twitter is sketched to establish whether we can
2094
IJRITCC | August 2014, Available @ http://www.ijritcc.org
_______________________________________________________________________________________
International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169
Volume: 2 Issue: 8 2094 – 2097
_______________________________________________________________________________________________
extract valid information from it. Proposing an event that necessarily valid words themselves (but rather common sub-
monitors tweet messages and delivers the warning from the strings, as the "earth" in "earthquk" and in "earthquake". In
explore. order to stem a word the algorithm tries to match it with
First, dragging numerous tweets related to target stems from the database, applying various constraints, such
events; second, extracting events from those tweets and as on the relative length of the candidate stem within the
estimate locations of events by using SVM; finally, develop word, for example, the short prefix "eq", which is the stem
a reporting system that extracts earthquakes from Twitter of such words as "EartQuk", "Quake" and "earth", would not
and sends a message to registered users. be considered as the stem of the word "earthquarter".
This paper is ordered as follows. Section 2 deals
about the Emergent Alert System. Section 3 focuses the 3.2 Support Vector Machines
ways of Algorithm used. Section 4 discusses about the
related works and discussion. Section 5 discusses about the
Support Vector Machines (SVMs, also support
future enhancements of NIA. Section 6 reviews the
vector networks) are supervised learning models with
conclusion of this paper. associated learning algorithms that analyze data and
2. EMERGENT ALERT SYSTEM recognize patterns. It used for classification and regression
analysis[9]. Given a set of training examples, each marked
Based on the previous research we are taking, as belonging to one of two categories, an SVM training
earthquake as our target. Why we have choose earthquake as algorithm builds a model that assigns new examples into one
a target because of the following reasons: category or the other, making it a non-probabilistic binary
linear classifier. An SVM model is a representation of the
1. It is very easy and accurately validates the event by event
examples as points in space, mapped so that the examples of
detection methodology.
the separate categories are divided by a clear gap that is as
2. Its gives out the meaningful value by detecting the event. wide as possible. New examples are then mapped into that
same space and predicted to belong to a category based on
Here, explaining our methods using an earthquake which side of the gap they fall on.
as a target event. First, to obtain tweets on the target event, The tweets are truly explains about the target. But
we apply semantic analysis of a tweet. For example, users they are not the original report of the event. It must confirm
might make tweets such as “Earthquake!” or “Now it is whether a given tweet is truly mentioning about the real
shaking,” in which earthquake or shaking could be target. For this confirmation we are using machine-learning
keywords, but users might also make tweets such as “I am
algorithms. To classify tweet content into positive or
attending an Earthquake Conference,” or “Someone is negative examples, we produce a model to classify the
shaking hands with my boss.” We prepare the training data examples.
and devise a classifier using a Support Vector Machine “Here earthquake, be safe”
(SVM) based on features such as keywords in a tweet, the
number of words, and the context of target-event words.
TABLE 1.SVM Example
After this classifier we find the content is informative or non
informative by using Naïve-Bayesian filtering method. Thus
Classification name classification
event-detection problem can be reduced to one of object
1 4 words,2nd word
detection and location estimation. A user has a mobile
2 Here,earthquake,be,safe
device. Through infrared communication or a Wi-Fi signal
of a mobile, the user location is estimated as providing 3 Here(finding of location)
location-based services such as navigation and museum
guides [9], [10]. Finally an emergent message has been sent We create three groups of classification for the above
to the registered user as well as the followers. example.
1-the number of words in a tweet messages ant the position
3. ALGORITHMS of keyword in a tweet.
3.1 Stemming: 2-the words in a tweet.
3-the words before and after the keyword.
Stemming is used to improve retrieval effectiveness The analysis of tweet is done using stop word elimination
and to reduce the size of indexing files. Several approaches and stemming which separates the sentences into a single set
to stemming are described--table lookup, affix removal, of words.
successor variety, and n-gram.
To find the useful tweets from the messages or 3.3 SVM is good for Classifiers
content, the tweet might include the target event. Here the High dimensional input space: Learning text classifiers, it
target event is earthquake. So we have to extract the content deals with many features. Since SVMs good for protection,
or messages with the keyword ’earthquake’. For example which does not necessarily depend on the number of
users might make tweets like “”. The keywords mentioned features, they have the potential to handle these large feature
are relates to the targeted keyword are confirmed by the spaces.
matching algorithms of stemming. Such algorithms use a
stem database, for example a set of documents that contain Few irrelevant features: in text categorization there are
stem words. These stems, as mentioned above, are not only very few irrelevant features. All features are ranked
according to their information gain. Then a naive Bayes
2095
IJRITCC | August 2014, Available @ http://www.ijritcc.org
_______________________________________________________________________________________
International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169
Volume: 2 Issue: 8 2094 – 2097
_______________________________________________________________________________________________
classifer [2] is trained using only those features ranked 1- interactions within Twitter reveal that the driver of usage is
200, 201-500, 501 -1000, 1001-2000, 2001-4000, 4001- a sparse and hidden network of connections underlying the
9962. A classifier using only that \worst” features have a declared set of friends and followers. Social networking of
performance much better to the conjecture that a good micro blogging services of twitter is used here. Sending of
classifier should combine many features (learn a \dense" messages to the followers either in a short way of
concept) and that aggressive feature selection may result in a communication.
loss of information.
Contains only few entries which are not zero. Kivinen et al. Joachims,[9] proposes the use of Support Vector
[4] give both theoretical and empirical evidence for the Machines for learning text classifiers. It analyzes the
mistake bound model that \additive" algorithms, which have particular properties of learning with text data and identities.
a similar inductive bias like SVMs, are well suited for SVMs achieve substantial improvements over the currently
problems with dense concepts and sparse instances. best performing methods and behave robustly over a variety
of different learning tasks. Explores key techniques for
Most text categorization problems are linearly handling and organizing text data. It identifies the benefits
separable: All Ohsumed categories are linearly separable of Support Vector Machines for text categorization. SVMs
and so are many of the Reuters tasks. The idea of SVMs is are a new learning method and very open to theoretical
to and such linear Separators. understanding and analysis. The properties of text related to
SVMs are very well suited for learning in this setting. SVMs
3.4 Naïve-Bayesian Filtering show substantial performance gains. the conventional text
classification methods SVMs will prove to be very robust,
We use probalistic models for the detection of the eliminating the need for expensive parameter tuning
target. Naive Bayes classifiers can be trained very efficiently
in a supervised learning setting. In many practical Kerstin Borau[10] establish the chance of actively
applications, parameter estimation for naive Bayes models produce language and the chance to use English as tool of
uses the method of maximum likelihood; in other words, one communication. It describes how to use the twitter with
can work with the naive Bayes model without accepting student and analyze the student message.Explains how
Bayesian probability or using any Bayesian methods. Twitter was used to learn specific aspects of language. Our
main finding is that it is suitable to train communicative and
A naive bayes classifier is a simple probabilistic classifier culture competence anytime anywhere with-out face-to-face
based on applying bayes theorem with strong independence interaction. Provides how to use twitter for the
assumptions. Probability model would be independent communication and the analyze of the text entry based on
model. the semantic usage.
P(C|F1,F2,F3......FN) over a dependent class
variable C with small number of outcomes or classes, Stefania Montani[11] proposes to describe a novel
conditional on several feature variables F1 through FN. The framework, in which time-series features are summarized by
problem is that if the number of features N is large or when means of Abstractions, and then retrieved resorting. Our
a when a feature can take on a large number of values, then approach grants for interpretability of the output results, and
basing such a model on probability tables is infeasible. understandability of retrieval process. It is domain-
Therefore reformulate the model to make it more tractable, independent. These of TA provide an easily interpretable
output, also for end users. Expected time of completion,
considering all types of inconsistency should be included to
P(C|F) = P(C) P(F|C) / P(F) prevent starvation for some tasks. Framework of time series
features are retrieved summarized and sorted for the
earthquake reporting system. Time of completion of the
expected event notification.
First we examine the actual data. It represents the
quality of target event, an earthquake. It is evident that Various researches have been studied using the
projection occurs in the number of tweets. Each corresponds informative messages from the twitter and presented as
to the target occurrence.concerning earthquake more probabilistic model. The motivation of our study is that the
earthquakes are occurred. twitter data is more time-sensitive and examines out the time
Chi-square is useful technique for finding whether events such as earthquakes.
the given data is representative of particular distribution.
4. RELATED WORKS 5. FUTURE ENHANCEMENTS

Best of the recent social media is Twitter. Many studies are undertaken to monitor the social media,
Plenty of researches are done in twitter. Regarding similar twitter. Most are aimed at the changes in the social state.
research are presented in this paper. Bernardo A. Huberman Our study is to use the twitter for real-time event detection.
[3]proposed a method to describe massive online social It is possible that many events can be finding by noticing
networks as a representation of social interactions that can twitter.
be used to study the propagation of ideas, social bond
dynamics and viral marketing, among others. Social
2096
IJRITCC | August 2014, Available @ http://www.ijritcc.org
_______________________________________________________________________________________
International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169
Volume: 2 Issue: 8 2094 – 2097
_______________________________________________________________________________________________
Expanding the system to detect various kinds of [9] Joachims, Baroper Str. Dortmund “Text
events using twitter. The system includes the assumption for Categorization with Support Vector Machines:
a single instance. We can detect the event like typhoon, Learning with Many Relevant Features” Thorsten
rainfall and even traffic jams. To realize more events at a University at DortmundInformatik LS8, Germany
time advanced probabilistic models are produced. That helps [10] Kerstin Borau, Carsten Ullrich, Jinjin Feng, and
to detect multiple events in a time. Ruimin Shen “Microblogging for language learning
:Using twitter to train communicative and cultural
A search query is important for seeking tweet .we
competence” Distance Education College of Shanghai
can monitor tweet to detect about the diseases or any other
Jiao Tong University.
medical related work. Advance algorithms can be useful for
[11] Stefania Montani, Giorgio Leonardi, Alessio
our future work.
Bottrighi, Luigi Portinale, and Paolo Terenziani
6. CONCLUSION ,(2013)“Supporting Flexible, Efficient, and User-
Interpretable Retrieval of Similar Time Series”, IEEE
As described in the paper .event detection of
transactions
earthquake is examined. Probabilistic models of SVMs are
[12] Takeshi Sakaki, Makoto Okazaki, and Yutaka
applied to find the informative and non-informative classes. Matsuo.(2013) “Tweet Analysis for Real-Time Event
Based on this we find an event. As an application the event Detection and Earthquake Reporting System
is send as a message to the registered users in a micro
Development”,IEEE
blogging services.
Twitter is used here as a micro blogging social
media and used in solving the real time problem of disasters.
It is hope that it provides some future integration in the
naturally inspired events.

REFERENCES
[1] M. Sarah, C. Abdur, H. Gregor, L. Ben, and M.
Roger, “Twitter and the Micro-Messaging
Revolution,” technical report, O’Reilly Radar, 2008. She completed B.TECH –Information Technology from
[2] A. Java, X. Song, T. Finin, and B. Tseng, “Why We Arulmigu Meenakshi Amman College of Engineering,
Twitter: Understanding Microblogging Usage and University of Madras in the year 2014 and also She
Communities,” Proc. Ninth WebKDD and First SNA- obtained her PG Degree in Computer Science from
KDD Workshop Web Mining and Social Network Thiruvalluvar College of Engineering and Technology,
Analysis (WebKDD/SNA-KDD ’07), pp. 56-65, Anna University in the year 2014.
2007.
[3] B. Huberman, D. Romero, and F. Wu, “Social
Networks that Matter: Twitter Under the
Microscope,” ArXiv E-Prints, http://
arxiv.org/abs/0812.1045, Dec. 2008.
[4] H. Kwak, C. Lee, H. Park, and S. Moon, “What is
Twitter, A Social Network or A News Media?” Proc.
19th Int’l Conf. World Wide Web (WWW ’10), pp.
591-600, 2010.
[5] G.L. Danah Boyd and S. Golder, “Tweet, Tweet,
Retweet:Conversational Aspects of Retweeting on
Twitter,” Proc. 43rd Hawaii Int’l Conf. System
Sciences (HICSS-43), 2010.
[6] A. Tumasjan, T.O. Sprenger, P.G. Sandner, and I.M.
Welpe,“Predicting Elections with Twitter: What 140
Characters Reveal About Political Sentiment,” Proc.
Fourth Int’l AAAI Conf. Weblogs and Social Media
(ICWSM), 2010.
[7] P. Galagan, “Twitter as a Learning Tool. Really,”
ASTD Learning Circuits, p. 13, 2009.
[8] K. Borau, C. Ullrich, J. Feng, and R. Shen,
“Microblogging for Language Learning: Using
Twitter to Train Communicative and Cultural
Competence,” Proc. Eighth Int’l Conf. Advances in
Web Based Learning (ICWL ’09), pp. 78-87, 2009.

2097
IJRITCC | August 2014, Available @ http://www.ijritcc.org
_______________________________________________________________________________________